FINAL REPORT INTERNATIONAL RESEARCH COLLABORATION AND SCIENTIFIC PUBLICATION
Activity-Based Travel Demand Model Systems Development for Banda Aceh City
Year 1 of 3 years planning
PRINCIPAL INVESTIGATOR Dr. Renni Angraini (NIDN 0023097101) Prof. H.J.P. Timmermans
SYIAH KUALA UNIVERSITY - INDONESIA EINDHOVEN UNIVERSITY OF TECHNOLOGY – THE NETHERLANDS November 2013
APPROVAL PAGE
Title
: Activity-based Travel Demand Model Development for Banda Aceh City
Principal Investigator Full Name NIDN Position Major of study Mobile Phone Number Email Address Member (1) Full Name NIDN Name of University Member (2) Full Name NIDN Name of University Collaborator Institution Name of Collaborator Address Responsible Person
: Dr. Renni Angraini, ST, M.Eng : 0023097101 : Lektor : Civil Engineering : 081263120232 :
[email protected] : Dr. Ir. Sofyan M.Saleh, M.Sc.Eng : 0012055903 : Syiah Kuala University : Dr. Ir. M.Isya, MT : 0011046208 : Syiah Kuala University : Urban Planning Group, Faculty of Architecture, Building & Planning, Eindhoven University of Technology, The Netherlands : Den Dolech 2, Eindhoven : Professor H. J. P. Timmermans
Year Implementation : Year 1 of 3 years plan Approved Budget for Year 1 : Rp. 150.000.000 Total Proposed Budget : Rp. 529.530.000 Banda Aceh, November 28, 2013 Approved by: Syiah Kuala University, Institute for Research,
Prof. Dr. Ir. Hasanuddin, M.S NIP: 19601114 198603 1 001
Team Leader,
Dr. Renni Angraini, ST, M.Eng NIP. 19710923 199702 2 001
2
TABLE OF CONTENTS I.
II.
INTRODUCTION ------------------------------------------------------------------- 1 1.1
Background ----------------------------------------------------------------- 1
1.2
Specific Purpose of Study ---------------------------------------------- 1
1.3
Urgency and Potential Results ---------------------------------------- 2
LITERATURE REVIEW ----------------------------------------------------------- 3 2.1
State-of-the-art ------------------------------------------------------------- 3
2.2
Conceptual Considerations --------------------------------------------- 5
2.2.1 Decision Making ---------------------------------------------------------- 5 2.2.2 Choice Behavior ---------------------------------------------------------- 7 2.2.3 Activity-based Approach -----------------------------------------------
8
2.2.4 Household Activity-Travel Behavior ---------------------------------
8
2.3
Decision Tree Induction ------------------------------------------------
9
2.3.1 Discrete Choices ---------------------------------------------------------
9
2.3.2 Continuous Choice ------------------------------------------------------ 10 2.3.3 Goodness-of-fit Measures --------------------------------------------- 12
2.4
A. Goodness-of-fit for Discrete Choices ---------------------------
12
B. Goodness-of-fit for Continuous Choices -----------------------
13
Uncertainty Analysis ----------------------------------------------------
14
III.
OBJECTIVES AND PURPOSE OF RESEARCH -------------------------- 15
IV.
METHODOLOGY ------------------------------------------------------------------- 17
V.
ACTIVITIES AND ACHIEVEMENTS ------------------------------------------ 22 5.1 ACTIVITIES --------------------------------------------------------------------- 22 5.2 ACHIEVEMENTS -------------------------------------------------------------- 23
VI.
FUTURE PLAN --------------------------------------------------------------------- 24
VII.
CONCLUSIONS AND RECOMMENDATION ------------------------------- 25
3
Chapter I. INTRODUCTION
1.1
Background
Transportation infrastructure development, in particular for public transport, is undeniable. The new development will impact the existing land use, and this is very costly. In addition, the environment in which transport analysis and infrastructure planning takes place has changed dramatically during the last few years. The focus is now on how to transform the transport system into a direction that could be sustainable in the long run, rather than on planning for infrastructure investment to meet new demand. For that reason, managing existing travel demand is necessary. By acknowledging individual travel behavior, transport policy can be applied to reduce traffic congestion. As a new focus on transport demand management, transport demand modeling has recently favored the activity-based approach to examine individuals’ travel behavior. Conventional transport demand models have often been criticized for being more suitable for construction of new infrastructure instead of managing existing travel demand, but lack the mechanism to support modern transportation policies. Consequently, in the United States for example, increasingly more planning authorities have are in the process of replacing traditional four-step models with activity-based models of travel demand. Other Asian countries are in a similar process. For some decades now, transport researchers have put considerable effort into developing what is called activity-based approaches to transport demand analysis. The basic idea is that travel demand is a derived demand based on people’s desire to take part in different activities. In particular, the interrelationships among different activities with respect to temporal and spatial constraints receive special attention. It means that such models treat the activities and the travel of households with respect to where and when the activities can be carried out, and how they may be scheduled, given characteristics of the households and potential opportunities, the transport networks and various institutional constraints. These mechanisms make these models more sensitive to recent policies, while in addition activity-based model do not have the lack of integrity, hampering four-step models, which by definition produce inconsistent results.
1.2
Specific Purpose of Study
4
The traditional four-step modeling is still a frontier in transportation research in Indonesia. As the models lacked any explanation in terms of human decision making and behavioral response to transport policy, it led to activity-based approach, which views travel as a result of people organizing their activities in time and space. Activity-based travel demand models have become an extremely relevant method for predicting travel demand and application of transport policy. Activity-based models are founded in behavioral theory and focus on the interdependencies between activity generation, transport mode choice, destination, stop pattern and route choice, in the context of multiple constraints that limit the choices of individuals and households. Hence, activity-based modeling is better suited to current transportation planning interests. In addition to that, this method has still rarely been used in Indonesian context. Based on this fact, the purposes of our study for the first year project are as follows: •
To distribute questionnaire about activity and travel of individual and household to Banda Aceh citizens
•
To develop the sequential model of activity-based transport demand modeling based on Albatross system by using individual and household data of Banda Aceh city
1.3
Urgency and potential results
Very few studies have been reported about an activity-based approach in Indonesian transportation research. Therefore, in this work, the operational system of transport demand based on Albatross system will be developed as a benchmark of activity-based travel demand modeling based on Indonesia population. This research will be conducted collaboratively by expert researchers about transport demand and travel behavior research from Syiah Kuala University, Banda Aceh Indonesia and Eindhoven University of Technology. Involvement of both teams in this research is expected to provide an optimal research quality, thus, it may contribute to solve the problem of transportation in Banda Aceh in particular and other cities in Indonesia in general. Furthermore, the expected output in this joint research is to continue the establishment of intensive cooperation between Syiah Kuala University and Eindhoven University of Technology in order to improve implementation of joint international seminars, publication of research results in international forum, etc. Thus, it will increase the academic achievement both of researcher from the University of Syiah Kuala and Eindhoven University of Technology in order to share scientific information between Netherland and Indonesia. 5
Chapter II.
LITERATURE REVIEWS
This section highlights state-of-the-art of feature descriptions and its applications in terms of decision making, choice behavior and activity-based travel demand modeling. We will present what has been studied, and also where the weaknesses, gaps, or areas needs further study. 2.1
State-of-the-Art
The choice of an activity-travel pattern that meets space–time, institutional and household constraints and satisfies as much as possible preferences of the individual, is an inherently complex cognitive task. Acknowledging this complexity, several authors have stressed the importance of understanding the decision making processes that underlie activity patterns and proposed computational process models that are consistent with cognitive theories of problem solving, decision making and learning. The production system is the most influential model of higher-order cognitive processes in cognitive sciences, since it was introduced by Newell and Simon in the early seventies (1972). A production system consists of a set of condition–action rules, called productions, representing long-term memory of the individual and a set of currently active facts or beliefs about a given problem in short-term memory. The systems describe problem solving as a cyclic process of matching productions with items in short-term memory and adding the action of an executed production to short-term memory (Anderson, 1983). As argued by Garling (1998), this cognitive theory points at forms of imperfect choice behavior not covered by current utility-maximization models based on economic theory. The implications are particularly relevant for activity-based models which consider choices on many facets in interaction. The number of possible activity-travel patterns is the product of all feasible activity sequences and activity profiles. As described by production systems, individuals search a solution space only partially based on heuristics that often yield satisfactory outcomes, but which are not necessarily optimal. The heuristics used determine the sequence in which decisions are made and the choice alternatives and attributes evaluated, and, hence, have impact on outcomes of the process. It is the central argument for computational process models that, even if one is interested in outcomes only, the process by which outcomes are generated needs to be represented in activity-based models. Garling et al. (1994) comprehensively discuss the theoretical underpinning of these models.
6
Despite the well-established status of computational process theory, existing attempts to develop computational process models (CPM) are limited. To review work in this field, we make a distinction between a weak and strong definition of CPM. A weak CPM applies a heuristic in the form of some sequential or partially sequential decision making process, but still assumes utility-maximization or some other form of unbounded rationality at the level of individual decision steps. Regardless whether or not reference is made to theories of heuristic search, several operational activity-based models exist that meet the weak definition. Examples are PCATS (Kitamura and Fujii, 1998), the model system proposed by Bhat (Bhat, 1999), STARCHILD (Recker et al., 1986a,b) and SMASH (Ettema et al., 1993). These models differ in various respects, but have in common that logit models or other forms of algebraic equations are used to predict single facet choices. Computational process models that fit the strong definition use a production system or some other rule-based formalism also at the level of individual choice facets. Models that meet this definition are scarce. The best example is SCHEDULER, developed by Garling et al. (1989). These authors developed a conceptual framework for understanding the process by which individuals organize their activities (Garling et al., 1989, 1992a,b). This framework has been applied to study the possible impact of the introduction of tele-commuting on the activity patterns of commuters (Garling et al., 1994). Only recently, parts of this conceptual model were elaborated and subject of experimental investigation (Garling et al., 1997, 1998, 1999). Strongly based on this model is GISICAS which uses search heuristics in combination with GIS to generate feasible schedules (Kwan, 1997). Another attempt to formulate a rule-based model has been reported in Vause (1997), but apparently this work has been discontinued. This brief summary of the state-of-the-art of computational process models indicates that, to the best of our knowledge, fully operational rule-based models have not been developed to date. Arguably, the lack of a method of empirically deriving rules of a production system has hampered progress in this field (Golledge et al., 1994). Furthermore, the verification of completeness and consistency of production systems presents a problem (Vause, 1997). While being consistent with assumptions of (strong) computational process models, the decision tree provides an alternative formalism that has favorable properties in both respects. Methods to induce decision trees from data are available from work in statistics and artificial intelligence. These methods develop a tree by recursively splitting a sample of observations into increasingly
7
homogeneous groups in terms of a given response variable. Decision trees induced in this way describe the data and, at the same time, meet requirements of completeness and consistency. A fully operational computational process model based on the decision tree formalism, Albatross (A Learning-based Transportation Oriented Simulation System), as the model is called, was developed for the Dutch Ministry of Transportation, Public Works and Water Management (Arentze and Timmermans, 1997). It was developed to explore possibilities of a rule-based approach and develop a travel demand model for policies impact analysis based on Netherland population. It has been developed for several phases underlying individual travel behavior from the first to four versions. In the fifth version it has been growing to include household decision making in the system by incorporating some choice facets such as task allocation, resource allocation and activity participation (Anggraini, 2009). 2.2. Conceptual considerations Consistent with the activity-based approach, we postulate that observed transportation patterns are the result of a complex decision-making process by which individuals try to achieve particular goals in the pursuit of their activities within the spatial-temporal and institutional constraints set by the environment. This section describes the concepts and assumptions underlying the Albatross model regarding decision making and choice behavior. 2.2.1. Decision making The organization of activities and related travel takes place in the context of an ever-changing physical environment, an uncertain transportation environment and multi-day variations in planned and unplanned activities that need to be completed. It is postulated that activity participation, allocation and implementation fundamentally take place at the level of the household. It is at that level that particular activities need to be performed, and it is also the household that is involved in the decision which activities to conduct. The actual generation and execution of activity calendars, programs and schedules covers a multitude of time frames. First, long-term decisions made at the household level strongly influence the generation and composition of activity calendars. Decisions regarding marital status, number of children, and the like, are irreversible or require years to change, and hence have a strong impact on the number and kinds of activities that need to be performed and the constraints that households face. These variables also influence the discretionary activities, reflecting an assumed relationship between socio-demographic variables and lifestyle. Other long-term decisions, such 8
as choice of residence, choice of work and workplace, and purchase of transport modes can in principle be changed in the short run, but in general represent the kind of choices that are not changed immediately. Hence, these decisions exert a strong influence on possible activity patterns as the location of the residence and workplace vis-_a-vis the transportation system represent the main locations of an activity pattern and are the cornerstones of decisions. Thus, these long-term decisions will influence household activity participation decisions. It is up to household members to allocate these activities to household members. The actual allocation will reflect task allocation mechanisms within the household, which will depend on genderspecific roles, and time pressures. The task allocation and related allocation of activities involves the set of activities that needs to be completed within a particular time horizon. It results in an individual activity program that is derived from the household activity calendar. We postulate that this process of program generation depends on the nature of the activities (mandatory versus discretionary), the urgency of completing a particular activity on a specific day as a function of the history of the activity scheduling and implementation process, and the desire to meet particular activity and time-related objectives. Once the individual activity program has been generated, the next step is to schedule these activities, which involves a set of interrelated decisions including the choice of location where to conduct a particular activity, the transport mode involved, the choice of other persons with whom to conduct the activities, the actual scheduling of activities contained in the activity program, and the choice of travel linkages which connect the activities in time and space. These activity scheduling decisions thus transform an individual’s activity program into an activity pattern, which is an ordered sequence of activities and related travel at particular locations, with particular start times and duration, with particular transport modes and perhaps coordinated with the activity patterns of other individuals. In this context, travel decisions represent a sub-decision. Transport mode decisions dictate the action space within which individuals can choose locations to conduct their activities. The organization of trips into chains allows individuals to conduct more activities within a specific time frame. The actual process of scheduling activities is conceptualized as a process in which an individual attempts to realize particular goals, given a variety of constraints that limit the number of feasible activity patterns. Several types of constraints can be identified:
9
(1)
Situational constraints impose that a person, transport mode and other schedule resources cannot be at different locations at the same time.
(2)
Institutional constraints, such as opening hours, influence the earliest and latest possible times to implement a particular activity.
(3)
Household constraints, such as bringing children to school, dictate when particular activities need to be performed and others cannot be performed.
(4)
Spatial constraints also have an impact in the sense that either particular activities cannot be performed at particular locations, or individuals have incomplete or incorrect information about the opportunities that particular locations may offer.
(5)
Time constraints limit the number of feasible activity patterns in the sense that activities do require some minimum duration and both the total amount of time and the amount of time for discretionary activities is limited.
(6)
Spatial-temporal constraints are critical in the sense that the specific interaction between an individual’s activity program, the individual’s cognitive space, the institutional context and the transportation environment may imply that an individual cannot be at a particular location at the right time to conduct a particular activity.
2.2.2. Choice behavior Having identified these constraints, the next question then is how individuals choose between feasible activity patterns. Unlike other models, which relied on utility-maximizing theory, we assume that choice behavior is based on rules that are formed and continuously adapted through learning while the individual is interacting with the environment (reinforcement learning) or communicating with others (social learning). Assume that an individual has just moved to a new city, which he does not know. To conduct his activities the individual will need to become involved in active search. Consider the choice of a location for shopping as an example. He may try locations at random, ask colleagues, consult newspapers or use some other strategy, but the result will be that he will visit a particular location for shopping. The experience with this location may not satisfy his expectations, in which case the individual will continue his active search behavior. He may, however, also be pleased with the experience. Having tried several locations, he will be able to compare the utilities associated with the different locations, and decide which location is the best under which 10
conditions (travel mode, specific activity, time of the day, day of the week, etc.). He may even induce from specific experiences the attributes of locations that co-vary with particular outcomes. In this way, associations between conditions and actions are formed. The complexity of condition–action associations may vary from individual to individual depending on the learning history. In particular, the choice between exploration and exploitation is a well-known dilemma in reinforcement learning theory (see Sutton and Barton, 1998). Risk takers may prolong exploration and accept the risk of negative outcomes to find choice alternatives that are more rewarding than the current best choice. Risk avoiders, on the other hand, may stop searching already in an early stage and accept the currently best choice while many alternatives have not been tried. As individuals may display different tendencies, we expect to find a wide variety of choice heuristics in terms of the extent to which context variables and choice alternative attributes are taken into account within a given population. Given some tendency to explore, refinement of condition–action associations will reflect the complexity of condition-reward contingencies in the environment. In dense urban areas and strong institutionalized environments, for example, we expect more differentiated behavior. Household setting, lifestyle and other factors determining complexity and time pressure on activities also have an impact. Moreover, associations tend to change over time. Existing associations may be weakened and new ones formed if the environment, preferences or life cycle of the individual changes. In sum, the learning theory on which Albatross is based implies that rules governing choice behavior are heuristic, context-dependent and adaptive in nature. 2.2.3
Activity-based Approach
The fundamental principle of the activity approach is that travel decisions are driven by a set of activities that form an agenda for participation and, as such, cannot be analyzed on an individual trip basis. Thus, the choice process associated with any specific travel decision can be understood and modeled only within the context of the entire agenda. Activity-based models are founded in behavioral theory and focus on the interdependencies between activity generation, transport mode choice, destination, stop pattern and route choice, in the context of multiple constraints that limit the choices of individuals and households. Activity-based models aim at predicting on a daily basis and for a household which activitievs are conducted, with whom, for how long, at what time, the location, and which transport mode is used when traveling is involved (Arentze and Timmermans, 2000, 2005, and Miller and Roorda
11
2003). McNally and Rindt (2008) characterized activity-based approaches generally reflect one or more of these factors: (1) travel is derived from the demand for activity participation; (2) the unit of analysis is sequence or pattern of behavior; (3) household and other social structures influence travel and activity behavior; (4) temporal, spatial, transportation, and interpersonal interdependencies constrain activity/travel behavior; and (5) activity-based approaches reflect the scheduling of activities in time and space. 2.2.4
Household activity-travel behavior
Activity-based approach to travel demand forecasting represents an attempt of improving the integrity of demand forecasting models by explicitly modeling various dependencies. These dependencies are not only concerned with the various choice facets (generation, destination, transport mode, etc), but also with dependencies between members of the household. The focus on the household as opposed to the traditional focus on the individual is especially important in the context of task and resource allocation and joint activities (Anggraini, et.al. 2008; Anggraini, 2009). The need to incorporate household decision making in activity-based approach has been acknowledged from the beginning. However, this topic has only recently received little attention, and a comprehensive model system at this level is still missing (Zhang et al., 2005, for a review paper). Several works have been made on the interactions of individuals within households (Gliebe and Koppelman, 2002, 2005; Scott and Kanaroglou, 2002; Srinivasan and Bhat, 2004), nevertheless, fewer attempts to integrate the interactions in activity-scheduling models. Activityscheduling models share an objective to predict the sequence of decisions that leads to an observed activity pattern of a household/individual. Incorporating mechanisms of household decision making should substantially improve the consistency and interdependencies in activity-travel decisions as an alternative to the more or less arbitrary breakdown of the multi-faceted decision problem, typical of the four-step models. However, although the degree of complexity and the sophistication of the econometric analysis have been substantially enhanced, at a more fundamental theoretical level considerably less has happened. Separating out the generation of activities and classifying certain patterns will at best allow us to capture only some aspects of how households cope with the constraints of their physical and social environments and organize their activities in time and space in an inherently dynamic context. In other words, a better understanding of this process and the underlying mechanisms and determinants is required. The focus on the household as opposed to the 12
traditional focus on the individual is especially important in the context of task and resource allocation and joint activities (Anggraini, 2009). 2.3
Decision Tree Induction Method
Every decision step in the process model is managed by decision tree. Each decision tree is derived from corresponding observations in the activity diary data set using a CHAID based induction method. This section considers the decision tree induction method used to determine decisions in the prediction stage, as explained in Arentze and Timmermans (2005). Discrete and continuous choices are separately discussed. 2.3.1 Discrete Choices The different levels at which decisions are to be made include the schedule, tour and activity level. Accordingly, the definition of a case differs between decision trees. As for example, the abstract illustration is assumed that at the given moment in the decision process, a decision is derived for N cases. A decision tree defines a classification function. Pr (k | Xj) = f(Xj)
(1)
where k is an index of leaf nodes of the tree and Xj is a vector of attribute levels for given case j. Since the type of decision trees used here is crisp (as opposed to fuzzy trees) and deterministic (as opposed to co-evolutionary trees), the probability of assigning case j to node k is one or zero. The action-assignment rule comes into operation after (1) and determines: Pr (i | k) = f(qk , δj)
(2)
where i is an index of discrete choice alternatives considered in the given decision tree, qk is the choice probability distribution across alternatives at the k-th node and δj is a zero-one vector indicating the availability of each choice alternative in case j. Note that, where qk is a characteristic of the decision tree, δj is to determined for each case in the prediction stage. The probability of selecting alternative i in case j is: Prj (i) =
∑
k
Pr(k | X j ) Pr(i | k )
(3)
Further, the probabilistic action-assignment rule f(qk , δj) used in ALBATROSS is specified. To simplify notations, the subset of cases assigned to leaf node k is considered and the subscript k in the symbols is dropped. The rule can be written as: 13
qi pij = δ ij ∑ δ ij qi i
∀i, j
(4)
where pij is the probability of selecting choice alternative i in case j (at leaf node k), δij is a zeroone variable indicating the availability of i in case j, and qi is the choice probability of alternative i dictated by the decision tree (at leaf node k) and estimated on the training set. As implied by this equation, probability pij is zero if i is not available and equals the second term on the RHS of the equation otherwise. 2.3.2
Continuous Choices
In the process model, continuous decision trees describe duration and start time choices. Rather than a choice probability distribution across discrete choice alternatives, these trees describe a specific distribution of the continuous duration or start time variable at each leaf node (Arentze and Timmermans, 2005). Thus, the continuous action-assignment equivalent of equation (2) defines the function: Pr (y | k) = f(Rk , Bj)
y = 0, 1, 2, ..., 1440
(5)
where Pr (y | k) is the probability of selecting value y at leaf node k, Rk is a vector of parameters defining the distribution at leaf node k and Bj is a set of tuples (b1, b2) defining unavailable or blocked ranges [b1, b2)] on dimension y in case j due to temporal constraints. Since times are measured in minutes and the schedule has a fixed time window (of 24 hours), y has a predefined minimum and maximum. Furthermore, we assume natural numbers for y. Continuous decision tree used in Albatross define distributions at each leaf node in terms of m – 1 cut-off points and the minimum and maximum of the range. The cut-off points divide the range into m intervals in such a way that an equal number of training cases at the leaf nodes is observed in each interval. As a consequence of this method, Rk specifies m+1 parameters. The number of elements of set Bj in a specific case is zero if the complete range is available and bigger than zero if parts of the range are blocked by constraints. To define the probabilistic continuous action-assignment rule used in Albatross, we the following symbols are used. Let Pj (y) denote the probability of selecting y = 1, …, 1440 in case j, m denote the number of equal-frequency intervals used in continuous decision trees, di represent the width of equal-frequency intervals used in continuous decision trees, di represent the width of equal frequency interval i, bij be the width of the blocked part of equal frequency interval i in 14
case j defined by the combination of Rk and Bj and Pj (y) =1, if y falls in the unblocked part of the interval i and 0 otherwise.
Pj ( y ) = ∑i Pr(i ) Pr( y | i )
∀j
(6)
where Pr (i) is the probability of selecting EFI (equal frequency interval) i and Pr (y | i) is the probability of selecting y given i. Pr (i) is defined as:
Pr(i ) =
1 d i − bij Cj m di
∀i
(7)
The first term represents the a-priori probability of selecting i. Because EFIs represent an equal number of cases, an equal number of cases, an equal probability is assumed for all m EFIs. The second and third terms define a correction this equal probability. The first correction is equal to the proportion of the available range in the EFI i and the third factor makes sure that probabilities sum up to one across EFIs. Similar as in the discrete case, we should include availability variables as potential predictor variables in inducing the tree to reduce the bias to the extent possible. 2.3.3
Goodness-of-Fit Measures
In order to measure the performance of the decision tree, different goodness-of-fit measures are used for discrete and continuous choice (Arentze and Timmermans, 2005). A.
Goodness-of-fit for Discrete Choices
There are two alternative goodness-of-fit measures for discrete choice decision trees. First, the so called likelihood or probabilistic theta is conceptualized as:
e = ∑k Pr( j − > k )∑i Pr(i | k ) Pr ' (i | k )
(8)
where e is the probability of correctly predicting the choice for any given case j in the same sample, Pr(j -> k) is the probability that j belongs to leaf node k, Pr(i | k) is the probability that choice i is observed in cases belonging to leaf node k and Pr’(i | k) is the probability of predicting i in those cases. The probabilities on the RHS of (28) can be found as:
Pr( j − > k ) =
fk n
(9) 15
Pr(i | k ) = Pr' (i | k ) =
f ik fk
(10)
where n is the total number of cases, fk is the number of cases at leaf node k and fik is the number of cases at leaf node k with observed choice i. The n and f variables all refer to the sample from which the tree was derived (i.e. the training set) so that the probabilities Pr(j -> k) and Pr(i | k) are to be interpreted as estimates of true probabilities for any sample of unseen cases. The predicted and observed probabilities in (8) are the same due to the probabilistic actionassignment rule used. Substituting (9) and (10) in (28) gives:
e = ∑k
f ik fk ∑ i n f k
2
(11)
And rewriting results in:
1 ∑ ( f ik ) e = ∑k i n fk
2
(12)
It should be noted that this measure assumes bias-free predictions. In reality, the actionassignment rule takes the availability of choice alternatives into account and therefore is more complicated than (10). The actual rule is given by (4). By comparing e with a null model, we can derive a measure of relative performance. We consider as the null-model a decision tree consisting of the root node only. Then, the likelihood or probabilistic theta for the null mode can be found as:
e0 = ∑i Pr(i ) Pr' (i ) 2
Or
1 f e 0 = ∑i i = 2 n n
(13)
∑ (f ) i
2
i
(14)
where fi is the overall frequency of choice i in the sample. The quotient
eincr =
e − e0 1 − e0
(15)
16
then indicates that the increase in likelihood as a ratio of the maximum increase that is possible given the null model. Note that this indicator is comparable to the (log) likelihood ratio commonly used as a measure of goodness-of-fit for conventional discrete choice models. The second measure is directly derived from the Chi-square statistic used as split criterion. The tree as a whole defines as I x K frequency table, where I is the number of choice alternatives and K is the number of leaf nodes. The Chi-square of this table can be taken as a measure of dependence between tree condition (leaf nodes) and choice. Because the value of Chi-square is dependent on sample size n, we use a standardization to obtain a measure on a (approximate) 0-1 scale, known as the contingency coefficient and defined as:
c=
χ2
(16)
χ2 +n
where χ 2 is Chi-square of the I x K frequency table and n is sample size. A zero value indicates a zero association between condition and choice and a value of one a maximum dependence. So, c can be interpreted as the discrete equivalent of the linear correlation coefficient between predictions and observations. B.
Goodness-of-fit for Continuous Choices
For continuous choice decision trees only one goodness-of-fit measure is available. This measure is directly derived from the F-statistic used as a split criterion. For the tree as a whole, the F-statistic is calculated as:
∑ F=
k
n k (m k − M ) 2 K −1
∑ /
k
n k (s k ) 2
n−K
(17)
where n is sample size, nk is number of cases at leaf node k, K is the number of leaf nodes, mk and sk are the mean and standard deviation of the distribution at leaf node k and M is the overall sample mean. Thus, F represents the ratio between between-group and within-group variance. The higher the value, the stronger the dependence between condition and action variable is. 2.4
Uncertainty analysis
Because Albatross is based on agent-based simulation, in order to separate policy effects form micro-simulation error it is important to run the model multiple times in scenario application. In more general terms, uncertainty analysis (Rasouli and Timmermans, 2012) is paramount to 17
assess the degree of uncertainty in the performance indicators generated by the model system for the different scenarios. This analysis will be part of the project. This part can rely on the results of the ongoing UncertWeb project which uses Albatross as an example of a complex model chains and which has developed software for such analysis.
18
Chapter III. OBJECTIVES AND PURPOSE OF RESEARCH
In overall, the research project aims to develop the operational system of activity-based travel demand modeling – Albatross - based on Banda Aceh city citizen data. As the current Albatross model system has been developed for the Dutch population, we intend to re-design it and apply it to the Indonesian population. Activity-based models have been widely used in European countries and in particular in the United States for recent years, but little research has been conducted in the context of Asia (although Albatross is currently being developed for South Korea). Therefore, it is a challenge to build this operational system as benchmark for Asian countries. The specific objectives are as follows: •
Developing the sequential model of activity-based transport demand modeling based on Albatross system by using individual and household data of Banda Aceh, in a form of SOFTWARE
•
Developing the synthetic population of other city in Aceh Province to compare with Banda Aceh data to analyze the sensitivity analysis
•
Developing the integrated model of Albatross system based on sequential choice model above.
•
Testing the validity of the integrated model of Albatross system and comparing it to another data population
•
Testing the sensitivity of the integrated model of Albatross system by applying the models to a particular scenario of change in Banda Aceh population. It is expected that the model will be more sensitive to such scenarios. The scenarios are such as increasing women labor participation, decreasing elder people participation, and increasing children participation in activity-travel
•
At least three international journal paper, three international conference, and six certified national journal papers can be submitted within three years period of research project.
•
Regarding international research collaboration, this research project is expected to conduct an intensive collaboration continuously for others authors and others activity.
The purpose of this research at least covers the following points: a. Solving transport problems and producing academic publications.
19
Our previous study also identified that various transport problems still exist in the area of activity-travel demand modeling. It will therefore lead us to the possibility of offering original contributions to the field, by means of producing more academic publications at international level. b. Fostering academic collaboration The implementation of this joint research will foster the mutual initiatives between Eindhoven University of Technology and Syiah Kuala University in advancing academic cooperation, particularly concerning joint research and publication activities. Academic achievements of both parties are expectedly going to be increased. Research Contributions The benefits of a comprehensiveness transport demand model are not restricted to prediction. It also means an increase in range and detail in information that can be generated and presented to the user for assessing the impact of scenarios. Furthermore, the systematic and better handling of household decision making implies that the model was expected to be more sensitive to scenarios that likely impact the household decision making. For long terms goal, it is expected to modeling the impacts of transport measures on emissions and energy consumption.
TRANSPORT MEASURES ON EMISSIONS AND ENERGY CONSUMPTION THE IMPACT
20
Chapter IV. METHODOLOGY
The research method for the first year is shown in Figure 1. The flowchart has been designed to illustrate step by step procedures that will be performed for the 1st year project starting by data collection. As this research primarily relies on individual and household activity and travel survey, the first year of research project only focused on collecting data. Each activity and the obtained results was recorded in a logbook and documented by computer software to ensure the accuracy of data. Several articles to be submitted to national journals, international conference and international journals have been drafted.
1)
Literature study The research was started with a comprehensive literature study by doing in-depth review on activity-travel behavior, including their applications in recent activity-based travel demand modeling.
2)
Data collection The data is required to estimate any activity-based model differ from the data necessary to calibrate conventional models. In particular, data on activity patterns are required in order to estimate and validate an activity-based model of transportation demand. The objective of a full activity-based model is to predict which activities are conducted where, when, for how long, with whom, and the involved of transport mode, and thus requires data on all these facets. Hence, as in common practice, it needs to collect activity diary data from household and individual to estimate activity-based models. Hence, the following data are required to be collected: - activity diary data of household and individual - the physical system: The physical system consists of the set of locations, where particular activity can be conducted. - the transportation system: The transportation system requires information about public transport attributes including route, travel times and costs. - the institutional context: The institutional context reflects to the opening hours, and hence it will be highly influential to the space-time prism that individual and household are facing. 21
3)
Report writing and conference/journal paper drafting At the end of research period, we will produce a comprehensive report that records all research activities and key findings. Scientific papers to be published in recognized international conferences and leading journals are also drafted.
Research Framework Activity-based models differ considerably from the traditional transportation forecasting methodologies in the sense that these models attempt to predict the interdependencies between the many facets of activity profiles. The difference with previous models is that the timing and scheduling of activities constitute the core of activity-based models. Consequently, the data required to estimate any activity-based model differ from the data necessary to calibrate conventional models. In particular, data on activity patterns are required in order to estimate and validate an activity-based model of transportation demand. Therefore, the activity diary data of household and individual is needed to be collected. In addition to the activity diary data, the estimation of the model requires space-time environment data consisting of the physical system, the transportation system, and the institutional context. The physical system consists of the set of locations, where particular activity can be conducted. The transportation system requires information about public transport attributes including route, travel times and costs. The institutional context reflects to the opening hours, and hence it will be highly influential to the space-time prism that individual and household are facing. For those reasons, the space-time environment data would be incorporated in the model, either as independent variables or as constraints, and it means that data about these factors need to be collected. For more detail, the framework is shown in Figure 1. Activity diary data and space-time environment data was possibly collected at the same time. Having prepared the questionnaire format and defined the sample among the population available, the questionnaire will be distributed to Banda Aceh city residents. By the time of collecting the activity-diary data, the space-time environment data is also collected. Note that, before collecting the actual data, the principal researcher from Syiah Kuala University came to the Netherlands to discuss about detail research plan, questionnaire designed, representative sampling and space-time environment data with researchers from Eindhoven University of Technology. This is important to do, the better preparation for data collection, the better results will be acquired. 22
START
Database Preparation
Space-time Environment data
Activity-diary data survey No Physical System
Transportation System
Institutional Context
Validity and Reliability
Yes
Compiling Data
Choice Facet Models Preparation
1st year Report
Writing papers for National and International Journal and Conference Figure 1 Flowchart Diagram for the 1st Year Research Project
Having collected those data, compiling data is necessary to do. Incomplete data will be diminished from database.
23
Survey Method Survey method is carried out by doing home interview survey to residents of Banda Aceh city. It covers 9 sub districts (kecamatan) in Banda Aceh city, i.e: Meuraxa, Jaya Baru, Banda Raya, Baiturrahman, Lueng Bata, Kuta Alam, Kuta Raja, Syiah Kuala, and Ulee Kareng. Home interview survey containing questionnaires will be performed to densely residential areas in those kecamatans,that is potential to generate/produce trip along with its activities. The questionnaire is dispatched to those households selected as samples. The survey is conducted on a regular basis to obtain travel and activity information of residents in Banda Aceh city. It is a household survey where data is collected of all household members for the diary day as well as general information about household and individual attributes such as, gender, age, vehicle ownership and driving license ownership, home location, individual income, occupation, number of working hours per week, etc. Respondents were also requested to give information about all trips made on a designated day as well as on the activities conducted on trip destinations. Information for each trip includes start time, trip purpose, destination, activity type at the destination, and transport mode. Situational variables are reported as well. All in all, this survey provides a comprehensive data source to analyze activity-travel behavior of Banda Aceh city residents.
Sample Size According to Stopher, et al (2008), sample size is probably the most controversial item in household travel surveys and one on which there is virtually no agreement, as evidenced by samples ranging from a few hundred to as much as 20,000 households. Procedures to develop suitable sample sizes are important issue, not only in the transportation field but also in the survey sampling literature. Obviously, there is such a wide variation in chosen sample sizes for household travel surveys arises from at least two issues: (1) available budget and (2) political rather than statistical justification of a particular sample size. The more data available, the better results will be acquired.
To determine the size of sample in a study, it uses a formula based on the proportion examined by Isaac & Michael. (Arikunto, 2006). Systematically, the amount of sample can be formulated as follows: S=
λ2 N P (1− P) d 2 ( N −1) + λ2 P (1− P)
24
where: S = number of samples; N = number of population; λ2 = chi-square; d = degree of accuracy, can be 1%, 5% and 10%; P = proportion of the population; Considering the population of Banda Aceh City and available budget from this research project, we decide to take sample of about 3,200 households for the first year project. It will be added more households for a following year. The population of Banda Aceh city in 2011 can be seen in Table 1.
Table.1. Population of Banda Aceh City in Year 2011 No. Kecamatan 1. Meuraxa 2. Jaya Baru 3. Banda Raya 4. Baiturrahman 5. Lueng Bata 6. Kuta Alam 7. Kuta Raja 8. Syiah Kuala 9. Ulee Kareng Jumlah Total Populasi
Total Population (person) 16.861 22.535 21.369 31.073 24.132 43.184 10.672 35.648 23.088 228.562
25
Chapter V. ACTIVITIES AND ACHIEVEMENTS
While the detailed activities are described in the research log book, our main activities in conducting the research can be listed as follows. −
Involving master students and undergraduate students in this research project
Recruitment of students was done as soon as announcement of getting funding from DIKTI released (around March 2013). There were 4 master students who are interested in household interview survey at the moment. They are interested in doing this survey because by collecting data together they can do different topics of thesis. This is the fact of powerful of this research, anyhow. The four master students have done the seminar proposals and all were accepted by supervisors and reviewers, by the end of June 2013. Currently we have 20 undergraduate students registered on this research. Hopefully, we can produce many potential bachelor and master students based on this research project. The theses of master students that closely related to this research project are as follows: 1. Trip Generation Models of Students in Banda Aceh City 2. Activity Generation Models of Workers and Non-Workers in Banda Aceh City 3. Joint Activity Participation Models based on Household Types 4. Vehicle Allocation Decisions in Vehicle-Deficient Households −
Visiting partner institution
The principal investigator from Syiah Kuala University visited the international collaborator at Eindhoven University of Technology (TUE), The Netherlands, from June 16 to June 21, 2013. Discussions about sampling, questionnaire design, method used, and data base conversion from Dutch data to Indonesian data were intensively conducted in The Netherlands. -
Data Collection
By the time of writing this final report, data collection has been distributed to Banda Aceh city residents involving about several people including students and covering about 3,000 households. In order to get more data, we need also to perform another data collection in the second year of project.
26
−
Presenting papers at international and national conferences.
Part of research findings are expected to be submitted to international and national conferences. Since the first year of research project has only deal with data collection, we haven’t done any publications elsewhere. Home interview survey needs extra effort to focus, in particular in terms of budget, time and members. Hopefully, we can produce a lot of publications in international and national conferences/journals, in particular based on students’ theses and final assignment. One potential article to be submitted to international conference is Individual and Household Travel Survey (Case Study Banda Aceh City) −
National and International Journal paper writing
Similar to conference publications, journal article has not done yet either at the moment. We had also discussed with TUE team, they agree that we have to focus on data collection for the first year. Only limited journals accept articles without any results. However, we have done several articles draft to be submitted to national and international journals. One article entitled Implementing the Activity-based Model of Albatross in Indonesian City (Case Study of Banda Aceh City) has been submitted to Journals of Intelligent Transport and Urban Planning. The comments have been received from two reviewers, at the moment. However, one of the reviewers suggested waiting for data analysis in order to get a better result. Other articles have been drafted to be submitted to either national or international journals and are as follows: 1. Model Bangkitan Pergerakan Pelajar dan Mahasiswa (Studi Kasus Kota Banda Aceh) 2. Model Bangkitan Aktivitas dari Pekerja dan Non-Pekerja pada Kota Banda Aceh 3. Analisis Keputusan Pengalokasian Kendaraan Pada Rumah Tangga yang Kekurangan Kendaraan 4. Modeling Within-Household Joint Activity Participation 5.2 Achievements As officially determined, the primary output of this research is publications in recognized international journals. Besides pursuing that goal, we also put efforts into producing papers and presented in international conferences to reach a broader academic audiences. However, we feel that the outcomes of research proposals from master students are also achievements. Based on students’ thesis we expect to submit several articles either to conferences or journals at national and international levels in a few months ahead.
27
Chapter VI. FUTURE PLAN
Having done the first year research project, we continue to prepare for the second year project. Since several articles have been drafted to be submitted to national - international journals and conference, however, primarily, data collection, data input and analysis have to be completed. Things to do for the first year project completion 1. Finalization of data collection In order to complete our work, we need to finalize the data collection consisting of home interview survey; the physical systems that consists of the set of locations, where particular activity can be conducted; the transportation system that requires information about public transport attributes including route, travel times and costs; and the institutional context that reflects to the opening hours 2. Inputting data Data is required to arrange and sorting, in order to get a better data base. The more detail inputting data, the better data will be acquired. 3. Finalization and submission of journal paper. Having done data terms as stated above, we will finalize the draft paper and submit it to a reputable national - international journals and conference. Things to do for the 2nd year project: 1. Adding more data in order to get bigger sample 2. Finishing articles and submit those for publications in national - international journals and conference 3. Attend national – international conferences 4. Visiting partner institution in Netherland to discuss again about the prototype of Albatross based on Indonesian population
28
Chapter VII. CONCLUSIONS AND RECOMMENDATION Based on our research activities, we conclude several points: 1.
By understanding activity-travel pattern of individual and households in Banda Aceh City, We can give recommendation to the local government in terms of transport planning and policy
2.
We are optimistic to develop this kind of operational system of travel demand as a benchmark of activity-based travel demand modeling for Indonesia and hence expected to be applicable to other Indonesian cities as well.
3.
Such a promising goal that we can assist the thesis of undergraduate and master students within couple of months under this research project, and hence a bunch of publications can be released either in conferences or journals at national or international levels.
4.
Fundamentally, we have not faced serious problems in conducting this research project. However, in order to get results faster, we suggest the funding should be dispatched as soon as the announcement was released, as we spend quite a lot of expenses by our own.
29
REFERENCES Anderson, N.H. (1982), Methods of Information Integration Theory, Academic Press, London. Anggraini, R., Arentze, T.A., and Timmermans, H.J.P., (2008). Paper presented at the 10th International Conference of Advanced Application in Transport and Technology, Athens, Greece Anggraini, R., (2009). Household Activity-Travel Behavior: Implementation of Within-Household Interactions. PhD dissertation. Eindhoven University of Technology, The Netherlands. Arentze, T.A. and H.J.P. Timmermans. (2000). ALBATROSS: A Learning-based Transportation Oriented Simulation System. EIRASS, Eindhoven University of Technology, The Netherlands. Arentze, T.A., and H.J.P. Timmermans. (2003). “Measuring Impacts of Condition Variables in Rule-Based Models of Space-Time Choice Behavior: Method and Empirical Illustration”. Geographical Analysis, 35, 24-45. Arentze, T.A. and H.J.P. Timmermans. (2004). “A learning-based transportation oriented simulation system”. Transportation Research Part B, 38, pp.613-633. Arentze, T.A. and H.J.P. Timmermans. (2005). ALBATROSS 2.0: A Learning-based Transportation Oriented Simulation System. EIRASS, Eindhoven University of Technology, The Netherlands. Ettema, D.F., Schwanen, T. & Timmermans, H.J.P. (2004), The Effect of Locational Factors on Task and Time Allocation in Households, Proceedings 83rd TRB Conference, Washington, D.C. (CD-Rom). Gliebe, J.P. and Koppelman, F.S. (2002). ”A Model of Joint Activity Participation between Household Members”. Transportation, 29, pp.49-72. Gliebe, J.P. and Koppelman, F.S. (2005). “Modeling Household Activity-Travel Interactions as Parallel Constrained Choices”. Transportation, 32, pp.449-471. Goulias, K.G. (2000). Companionship and Altruism in Daily Activity Time Allocation and Travel by Men and Women in the Same Households. In Proceeding of TRB 200, Washington, D.C., US. Kitamura, R. & Fujii, S. (1998), Two Computational Process Models of Activity-Travel Choice, In: T. Garling, T. Laitila & K. Westin (eds.): Theoretical Foundations of Travel Choice Modeling, Elsevier, Oxford, 251-279. Miller, E.J., and M.J. Roorda. (2003). “A Prototype Model of Household Activity/Travel Scheduling”. Proceedings of the 2003 Transportation Research Board, Washington DC, USA.
30
Rasouli, S. and H.J.P. Timmermans (2012), Uncertainty in travel demand forecasting models, Literature review and research agenda, Transportation Letters Recker, W.W. (1995). The Household Activity Pattern Problem: General Formulation and Solution, Transportation Research, 29B, 61-77. Scott, D. & Kanaroglou, P. (2002). An activity-episode generation model that captures interaction between household heads: development and empirical analysis. Transportation Research B, 36B: 875-896. Srinivasan, S. & Bhat, C. (2004). Modeling the generation and allocation of shopping activities in a household. Paper presented at the 83rd Annual Meeting of the Transportation Research Board, Washington, DC. Stopher, P.R., R. Alsnih, C.G. Wilmot, C. Stecher, J. Pratt, J. Zmud, W. Mix, M. Freedman, K. Axhausen, M. Lee-Gosselin, A.E. Pis arski, and W. Brög (2008). Standardized Procedures for Personal Travel Surveys, NCHRP Report 571, Transportation Research Board, Washington, DC, 103 pages. Timmermans, H.J.P., and R. van der Heijden (1987). Uncovering Spatial Decision-Making Processes: A Decision Net Approach Applied to Recreational Choice Behavior. Journal of Economic and Social Geography, 4, 297-304. Zhang, J., Timmermans, H.J.P., Borgers, A., 2005. A Model of Household Task Allocation and Time Use. Transportation Research Part B, 39, 81-95.
31
32
33
Article #1 (Draft article to be submitted to National Journal)
Model Bangkitan Pergerakan Pelajar dan Mahasiswa (Studi Kasus Kota Banda Aceh) Juanda1, Renni Anggraini2*, M. Isya2 Mahasiswa Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh 2 Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh *Email:
[email protected] 1
Abstract Penelitian ini bertujuan untuk mendapatkan model bangkitan pergerakan aktivitas pelajar sehari-hari dan mengidentifikasi faktor-faktor yang mempengaruhi terjadinya pergerakan oleh pelajar di sembilan Kecamatan di Kota Banda Aceh ketempat beraktivitas. Model yang diperoleh tersebut bisa dipergunakan untuk memprediksi bangkitan pergerakan yang terjadi di Kota Banda Aceh pada 5, 10 dan 15 tahun mendatang. Survei primer dilakukan melalui pengisian kuesioner pada 12.447 responden (3.112 kepala keluarga) yang bertempat tinggal di sembilan Kecamatan di Kota Banda Aceh. Hasil dari kuesioner ditabulasikan menjadi variabel bebas dan variabel terikat, kemudian dianalisis menggunakan program SPSS-17. Persamaan regresi digunakan untuk memodelkan bangkitan pergerakan pelajar berdasarkan aktivitas di Kecamatan tersebut. Hasil dari penelitian ini diharapkan diperoleh tiga kelompok model aktivitas pergerakan yaitu : aktivitas mandatory, aktivitas maintenance dan aktivitas discretionary yang dipengaruhi oleh faktor sosio ekonomi dan tata guna lahan di Kota Banda Aceh. Hasil dari penelitian ini juga diharapkan dapat memberikan masukan kepada pemerintah daerah dalam sistem transportasi di Kota Banda Aceh akibat terjadinya bangkitan dan tarikan pergerakan saat ini untuk memprediksi karakteristik dan besarnya tarikan pergerakan di masa mendatang. Kata kunci : Model bangkitan tarikan pergerakan, aktivitas mandatory, aktivitas discretionary.
I.
maintenance,
aktivitas
PENDAHULUAN Kota Banda Aceh saat ini terus mengalami perkembangan jumlah penduduk. Perkembangan ini
terlihat dari pesatnya pembangunan rumah dan rumah toko (ruko), sehingga menyebabkan meningkatnya aktivitas pergerakan individu yang terjadi di kawasan tersebut. Model kebutuhan pergerakan yang dilakukan oleh individu yang terjadi di suatu kawasan perlu diketahui besarnya tingkat kebutuhan akan pergerakan dengan mempelajari berbagai variasi hubungan antar ciri pergerakan. Dalam hal ini akan dilakukan kajian pergerakan yang terjadi di kota Banda Aceh di sembilan kecamatan yaitu Kecamatan Meuraxa, Jaya Baru, Banda Raya, Baiturrahman, Lueng Bata, Kuta Alam, Kuta Raja, Syiah Kuala, dan Ulee Kareng dengan menggunakan data rinci mengenai tingkat kebutuhan pergerakan aktivitas pelajar sehari-hari. Populasi Kawasan yang ditinjau pada Kota Banda Aceh adalah 34
sekitar 228.562 jiwa (Aceh Dalam Angka 2011) dengan keseluruhan total luas wilayah adalah 61.359 Ha. Berbagai jenis aktivitas yang berbeda-beda dilakukan setiap harinya oleh masing-masing penduduk, akibat pesatnya tingkat kemajuan Kota Banda Aceh menyebabkan banyaknya pergerakan individu yang dibangkitkan pada kawasan tersebut seperti aktivitas pergerakan pelajar. Aktivitas pelajar yang berada di kecamatan ini pada umumnya mempunyai aktivitas di sekitaran kota Banda Aceh seperti les, kuliah, kegiatan sosial, berbelanja, rekreasi dan sebagainya yang dapat mempengaruhi tingkat pelayanan jalan utama di kota Banda Aceh. Melihat hal tersebut, diperlukan penelitian tentang kebutuhan pergerakan seorang pelajar di kecamatan tersebut yang bertujuan untuk memodelkan kebutuhan pergerakan pelajar perhari pada setiap rumah tangga di kawasan tersebut. Adapun pergerakan pelajar yang akan di tinjau ada 4 kategori, yaitu pelajar sekolah dasar, sekolah menengah pertama, sekolah menengah atas, dan mahasiswa. Pelajar yang duduk di bangku sekolah dasar, sekolah menengah pertama, dan sekolah menengah atas memiliki jam masuk dan pulang sekolah yang rutin setiap harinya, tetapi setelah jam pulang sekolah tidak semua pelajar melakukan pergerakan yang sama, kecuali mahasiswa yang memiliki jam belajar yang tidak tetap dan untuk tinjauan mahasiswa memiliki batasan umur sampai dengan 30 tahun. Oleh karena itu, kategori pergerakan pelajar tersebut masing-masing memiliki perbedaan kegiatan. Hasil yang ingin dicapai pada penelitian yang akan dilakukan adalah untuk mengetahui faktorfaktor sosio ekonomi dan tata guna lahan yang sangat berpengaruh dan seberapa besar pengaruhnya terhadap pergerakan aktivitas pelajar di setiap rumah tangga di Kota Banda Aceh yang pada akhirnya diperoleh suatu model kebutuhan pergerakan yang mana nantinya model yang diperoleh tersebut bisa digunakan untuk memprediksikan kebutuhan akan pergerakan yang terjadi di kawasan tersebut dimasa yang akan datang. Meningkatnya perkembangan penduduk di kota Banda Aceh mengakibatkan bertambahnya jumlah pergerakan pelajar seperti sekolah, mengaji, olahraga, belanja, rekreasi, kunjungan sosial dan lainnya, yang akan mempengaruhi kapasitas jalan dan berdampak pada permasalahan pengembangan transportasi kota Banda Aceh yang merupakan tujuan aktivitas pelajar di Kecamatan Meuraxa, Jaya Baru, Banda Raya, Baiturrahman, Lueng Bata, Kuta Alam, Kuta Raja, Syiah Kuala, dan Ulee Kareng. Maka daripada itu dibutuhkan suatu permodelan kebutuhan pergerakan aktivitas pelajar di sembilan kecamatan tersebut. Tujuan penelitian ini adalah untuk mengindentifikasi faktor-faktor yang mempengaruhi terjadinya pergerakan aktivitas pelajar sehari-hari di kawasan tersebut dengan memprediksi dan melakukan pemodelan kegiatan pelajar yang dilakukan setiap hari. Manfaat dari penelitian ini nantinya dari hasil model yang diperoleh dapat dipergunakan untuk memprediksi bangkitan pergerakan yang terjadi di kota Banda Aceh pada 5, 10 dan 15 tahun mendatang. 35
Penelitian ini dilakukan dengan menggunakan kuesioner yang berisikan data rumah tangga dan data pergerakan individu rumah tangga. Dengan mempertimbangkan luasnya daerah kajian penelitian dan keterbatasan waktu dan biaya peneliti, maka digunakan batasan-batasan sebagai berikut : 1.
Perjalanan yang dilakukan pelajar didalam suatu keluarga (rumah tangga) yang hanya dianalisis berdasarkan home base trip, yaitu semua perjalanan yang berasal dari rumah dan diakhiri dengan pulang kerumah.
2.
Daerah penelitian diambil pada sembilan kecamatan yang berada di kota Banda Aceh. Pengumpulan data untuk keperluan analisa diperoleh dengan metode teknik sampling terhadap jumlah rumah tangga yang ada di sembilan kecamatan tersebut.
3.
Metode yang digunakan untuk mendapatkan model kebutuhan pergerakan adalah dengan menggunakan analisis Regresi Linier Berganda (Multiple Linear Regression Analysis) yang diolah dengan Software SPSS (Statistical Product and Service Solution). Bangkitan pergerakan adalah banyaknya lalu lintas yang ditimbulkan oleh suatu zona atau tata
guna lahan persatuan waktu (Tamin, 1997). Pembangkit perjalanan atau bangkitan perjalanan ini berhubungan dengan penentuan jumlah perjalanan keseluruhan yang di bangkitkan oleh suatu kawasan, Trip Generation terbagi atas dua bagian yaitu Trip Production dan Trip Attraction. Bangkitan dan tarikan pergerakan digunakan untuk menyatakan bangkitan pergerakan pada masa sekarang, yang akan digunakan untuk meramalkan pergerakan pada masa mendatang. Bangkitan pergerakan ini berhubungan dengan penentuan jumlah keseluruhan yang dibangkitkan oleh sebuah kawasan. Parameter tujuan perjalanan yang berpengaruh di dalam produksi perjalanan (Levinson, 1976), adalah: 1.
Tempat bekerja
2.
Kawasan perbelanjaan
3.
Kawasan pendidikan
4.
Kawasan usaha (bisnis)/sosial
5.
Kawasan hiburan (rekreasi) Model merupakan penyederhanaan dari keadaan sebenarnya dan model dapat memberikan
petunjuk dalam perencanaan transportasi. Karakteristik sistem transportasi untuk daerah-daerah terpilih seperti CBD (Central Bussiness District) sering dianalisis dengan model. Model memungkinkan untuk mendapatkan penilaian yang cepat terhadap alternatif-alternatif transportasi dalam suatu daerah (Morlok, 1991). Dalam pemodelan bangkitan pergerakan, metode analisis regresi linear berganda (Multiple Linear Regression Analysis) yang paling sering digunakan baik dengan data zona dan data rumah tangga atau 36
individu. Metode analisis regresi linear berganda digunakan untuk menghasilkan hubungan dalam bentuk numerik dan untuk melihat bagaimana variabel saling terkait. Ada beberapa asumsi statistik harus dipertimbangkan dalam menggunakan metode analisis regresi linear berganda, sebagai berikut: 1.
Variabel terikat (Y) merupakan fungsi linear dari variabel bebas (X).
2.
Variabel, terutama variabel bebas adalah tetap atau telah diukur tanpa galat.
3.
Tidak ada korelasi antara variabel bebas.
4.
Variansi dari variabel terikat terhadap garis regresi adalah sama untuk nilai semua variabel terikat.
5.
Nilai variabel terikat harus tersebar normal atau minimal mendekati normal. Analisis regresi linear berganda (Multiple Linear regression Analysis) yaitu suatu cara yang
dimungkinkan untuk melakukan beberapa proses iterasi. Yang pada akhirnya diperoleh persamaan model regresi linier berganda seperti di bawah ini : Y = a + b1X1 + b2X2 + ……… + bnXn Dimana : Y
= variabel terikat (jumlah produksi perjalanan),
a
= konstanta (angka yang akan dicari)
b1, b2, …,bn
= koefesien regresi (angka yang akan dicari)
X1, X2,…,Xn = variabel bebas (faktor-faktor berpengaharui)
Pengambilan sampel dilakukan untuk mendapatkan sampel dengan jumlah relatif kecil dibandingkan dengan jumlah populasi tetapi mampu mewakili seluruh populasi tersebut. Untuk itu sangat penting menentukan cara yang tepat dalam menarik sampel yang dimaksud agar benar-benar mampu mewaikili kondisi seluruh populasi. Untuk memudahkan dan menentukan besarnya ukuran sampel dalam suatu penelitian maka dapat digunakan rumus yang berdasarkan proporsi yang di kemukakan oleh Issac dan Michael. (Arikunto, 2006). Secara sistematis, besarnya sampel dari populasi dapat dirumuskan sebagai berikut :
λ 2 N P (1 − P) S= 2 d (N − 1) + λ 2 P (1 − P) Dimana : S
= Jumlah sampel
N
= Jumlah Populasi
λ2
= Harga tabel chi-kuadrat, dengan dk = 1
d
= Ketelitian (error), bisa 1%, 5% dan 10%. 37
P
= Proporsi dalam populasi
Didalam penelitian transportasi, di jelaskan bahwa perjalanan berasal dari rumah tangga yang dilakukan dibawah spasial dan kepentingan sementara. Menurut Anggraini, dkk (2006), kegiatan aktivitas dibagi ke dalam 3 (tiga) kelompok yaitu mandatory, maintenance dan discretionary. Aktivitas bekerja dan sekolah merupakan aktivitas compulsory atau mandatory yang dilakukan oleh individu perorangan. Aktivitas seperti kegiatan pelajar diluar jam sekolah, belanja harian, belanja yang tidak dilakukan perhari, membawa dan mengantar anak atau orang lain dan sebagainya merupakan aktivitas maintenance. Dan prinsip ini cukup jelas jika ada salah satu anggota keluarga yang melakukan aktivitas tersebut berarti termasuk obyek aktivitas maintenance. Sedangkan aktivitas seperti rekreasi, kunjungan sosial dan aktivitas senang-senang lainnya dikategorikan sebagai aktivitas discretionary dan dapat dilakukan bersama-sama atau tidak bersama-sama. Dalam kasus kegiatan pelajar, pola aktivitas perjalanan dianggap sebagai suatu aktivitas yang dilakukan pelajar dalam satu hari baik pada jam sekolah maupun di luar jam sekolah yang diselingi berbagai jenis aktivitas. Pola aktifitas pelajar harian ini tergantung pada aktivitas yang dilakukan selama satu hari di mulai berangkat dari rumah, pada jam sekolah maupun di luar jam sekolah, dan kembali ke rumah.
II.
METODOLOGI
2.1
Survei Pendahuluan Survei pendahuluan merupakan langkah awal yang dilakukan sebelum survei sesungguhnya
dilaksanakan yaitu untuk mendapatkan data lapangan yang sesuai dengan yang diharapkan, hal ini dimaksudkan untuk : 1.
Mengetahui keadaan lapangan
2.
Memilih lokasi yang paling cocok diantara beberapa lokasi yang telah direncanakan
3.
Menentukan desain sampel yang cocok dilaksanakan di lapangan
4.
Menetapkan strategi pelaksanaan survei Melalui survei pendahuluan diperoleh data-data sekunder yang akan digunakan untuk
memperoleh informasi daerah penelitian dan populasinya. Pengambilan data pada suatu penelitian dapat dilakukan dengan 2 metode yaitu : 1.
Metode Survei
2.
Metode Sampel
2.1.1 Metode survei 38
Metode survei yang digunakan metode survei wawancara rumah tangga yang telah dilakukan pada penelitian sebelumnya, dan dilakukan metode survei asal-tujuan untuk menetapkan titiktitik penelitian. Survei ini dilakukan pada kawasan-kawasan pemukiman yang sangat potensial menimbulkan perjalanan dengan berbagai aktivitasnya. Oleh karena itu, data yang diperoleh dari survei ini berguna sebagai input data untuk tahap bangkitan perjalanan, karena zona pemukimanlah yang memproduksi perjalanan. Objek survei ini adalah personil yang mendiami rumah-rumah di kawasan pemukiman yang ada di Kota Banda Aceh. Selanjutnya dianalisis karakteristik objek yang akan dijadikan variabel/faktor penyebab terproduksinya perjalanan dari dari zona pemukiman menuju ke tempat tujuan belajar. Variabel tersebut seperti jumlah pelajar, jumlah pekerja, jarak tempuh, jenis kendaraan, banyaknya anggota keluarga, dan karakteristik lain yang berhubungan. Adapun alat kelengkapan survei ini salah satunya adalah daftar pertanyaan yang formatnya telah ditentukan sebelumnya dan variabel yang disesuaikan kebutuhan.
2.1.2 Metode sampel
Metode ini mengumpulkan data dan informasi dengan mencatat sebagian kecil objek pengamatan yang merupakan bagian dari populasi secara keseluruhan dengan cara sampling. Sampling adalah pengumpulan data atau penelitian hanya pada elemen sampel (sebagian dari elemen populasi) yang diteliti.
2.2
Jenis dan Sumber Data
Jenis data yang digunakan dalam penelitian ini adalah data primer dan data sekunder. Data primer adalah data yang diperoleh lansung dari responden atau obyek yang diteliti, atau ada hubungannya dengan yang diteliti. Dalam penulisan ini data primer yang dimaksud adalah data yang sumbernya diperoleh langsung dari responden/penghuni perumahan dengan kuesioner, yaitu data jumlah anggota keluarga (orang), jumlah kegiatan pelajar perhari (orang), jumlah anak di bawah 6 tahun (orang), jumlah anak SD dalam keluarga (orang), jumlah anak SMP dalam keluarga (orang), jumlah anak SMA dalam keluarga (orang), jumlah anak kuliah dalam keluarga (orang), anggota keluarga yang bekerja (orang), jarak tempuh ke sekolah, moda yang digunakan, waktu berpergian, dan kepemilikan kendaraan.
39
Sedangkan data sekunder adalah data yang lebih dulu dikumpulkan dan dilaporkan oleh orang atau instansi diluar diri peneliti sendiri, walaupun yang dikumpulkan itu sesungguhnya data yang asli. Data sekunder diperoleh dari instansi-instansi terkait dan perpustakaan. Di samping itu melalui survey pendahuluan juga dapat diperoleh data-data sekunder yang akan digunakan untuk memperoleh informasi daerah penelitian dan populasinya.
2.3
Metode Pengambilan Sampel
Pengambilan sampel adalah mendapatkan sampel dengan jumlah relatif kecil dibandingkan dengan jumlah populasi tetapi mampu mempresentasikan seluruh populasi tersebut. Untuk itu sangat penting menentukan cara yang tepat dalam menarik sampel yang dimaksud agar benar-benar mampu mempresentasikan kondisi seluruh populasi. Mengingat karakteristik sosio ekonomi penduduk dan subjek yang terdapat pada setiap wilayah berbeda-beda, maka untuk memperoleh sampel yang representatif, pengambilan subjek dari setiap wilayah ditentukan seimbang atau sebanding dengan banyaknya subjek dalam masing-masing wilayah. Teknik penarikan sampel yang dipergunakan adalah Proportional Sample atau sampel proporsi atau sampel imbangan Sampel adalah sebagian atau wakil populasi yang diteliti. Dinamakan penelitian sampel apabila kita bermaksud untuk menggeneralisasikan hasil penelitian sampel. Yang dimaksud dengan menggeneralisasikan adalah mengangkat kesimpulan penelitian sebagai suatu yang berlaku bagi populasi. Penentuan jumlah sampel penelitian menggunakan teknik pengambilan sampel secara teknik simple random sampling yaitu bentuk sampling probabilitas yang sifatnya sederhana, dimana tiap sampel yang berukuran sama memiliki suatu probabilitas atau kesempatan yang sama untuk dipilih dari populasi. Alasan menggunakan teknik sampling adalah untuk mempermudah penulis dalam pengumpulan data melalui kuesioner yang dibagikan serta adanya keterbatasan dana dan waktu penelitian. Adapun cara pengambilan sampel yang peneliti lakukan dengan menggunakan rumus berikut ini :
S=
λ 2 N P (1 − P) d 2 (N − 1) + λ 2 P (1 − P)
Dimana : S
= Jumlah sampel
N
= Jumlah Populasi
λ2
= Harga tabel chi-kuadrat, dengan dk = 1
d
= Ketelitian (error), bisa 1%, 5% dan 10%.
P
= Proporsi dalam populasi 40
Dengan tingkat ketelitian (d) 5% dan tingkat kepercayaan 95%. Sedangkan nilai λ2 = 3,841 pada derajat kebebasan 1. Pada proporsi populasi dalam perhitungan P (1 – P) dimana diambil nilai P = 0,5. Dari ketentuan diatas maka diperoleh jumlah sampel yang dibutuhkan adalah :
S=
λ 2 N P (1 − P) d 2 (N −1) + λ 2 P (1 − P)
(3,841) 2 × (16.861) × 0,5 (1 − 0,5) S = = (0,05) 2 × (16.861 − 1) + (3,841) 2 0,5(1 − 0,5)
62.188,76 62.188,76 = 42,15 + 3,688 45,838
S = 1.356,70 responden Jika dalam satu kepala keluarga terdiri dari 4 orang, maka jumlah sampel dapat menjadi :
S=
1.356,70 = 339,17 KK 4 Jumlah keseluruhan sampel di Kota Banda Aceh dapat dilihat pada Tabel 3.2 di bawah ini :
No.
Kecamatan
Jumlah
Jumlah Sampel
Populasi
Responden
Jumlah Sampel KK
1
Meuraxa
16.861,00
1.356,70
339,17
2
Jaya Baru
22.535,00
1.384,73
346,18
3
Banda Raya
21.369,00
1.380,11
345,03
4
Baiturrahman
31.073,00
1.408,50
352,12
5
Lueng Bata
24.132,00
1.390,38
347,60
6
Kuta Alam
43.184,00
1.426,62
356,66
7
Kuta Raja
10.672,00
1.296,25
324,06
8
Syiah Kuala
35.648,00
1.416,73
354,18
9
Ulee Kareng
23.088,00
1.386,77
346,69
228.562,00
12.446,81
3.111,70
TOTAL
Keterangan
Banda Aceh
Tabel 3.2 : Jumlah Sampel Populasi di Kota Banda Aceh
III.
HASIL DAN PEMBAHASAN
Untuk menjawab perumusan masalah yang telah ditetapkan, yaitu berapa besar pengaruh variabel mengenai bangkitan pergerakan (X) seperti : jumlah anak di bawah 6 tahun (orang), jumlah anak SD dalam keluarga (orang), jumlah anak SMP dalam keluarga (orang), jumlah anak SMA dalam keluarga 41
(orang), jumlah anak kuliah dalam keluarga (orang), anggota keluarga yang bekerja (orang), jarak tempuh ke sekolah, moda yang digunakan, waktu berpergian, dan kepemilikan kendaraan, jumlah kegiatan pelajar perhari (Y), perlu dilakukan beberapa tahapan penting untuk menganalisis data yang diperoleh melalui survei kuesioner. Uji korelasi dan proses kalibrasi dilakukan dengan menggunakan bantuan software SPSS (Statistical Product and Service Solution) yaitu suatu program statistik yang mampu memproses data statistik secara tepat dan tepat serta menyajikannya dalam berbagai output yang dikehendaki para pengambil keputusan.
3.1.1 Pengolahan data
Pengolahan data dilakukan apabila data primer dan data sekunder telah terkumpul, maka data tersebut dilakukan pengolahan dengan langkah-langkah sebagai berikut: a. Editing, yaitu mengoreksi kesalahan-kesalahan dalam pengisian atau pengambilan data. Pada tahap ini data yang telah dikumpulkan dilakukan pemeriksaan ulang nama identitas responden, memeriksa kelengkapan data. Hal ini dilakukan untuk mencegah kekeliruan yang mungkin terjadi. b. Coding, yaitu mengklarifikasikan jawaban responden menurut macamnya dengan memberikan kode tertentu. Pada tahap ini data yang telah diperoleh diberikan angka-angka atau kode tertentu untuk memudahkan pengenalan data. c. Transfering, yaitu data yang telah diberi kode secara berurutan dari responden pertama sampai dengan responden terakhir, dan selanjutnya dimasukkan ke dalam tabel. Semua data primer dan data sekunder dilakukan analisis dengan menggunakan perangkat lunak Microsoft Excel dan SPSS version 17.
3.1.2 Analisis bivariat
Analisis bivariat, yaitu analisis uji korelasi untuk melihat hubungan antar variabel yaitu variabel terikat dengan variabel bebas. Variabel bebas harus mempunyai korelasi tinggi terhadap variabel terikat dan sesama variabel bebas tidak boleh saling berkorelasi. Apabila terdapat korelasi diantara variabel bebas, pilih salah satu yang mempunyai nilai korelasi yang terbesar untuk mewakili. Koefisien korelasi digunakan untuk mengetahui bagaimana keeratan hubungan antara satu variabel dengan variabel lainnya.
3.1.3 Analisis multivariat 42
Untuk data yang lebih dari dua variabel bebas dilakukan analisis multivariat yaitu untuk mencari pengaruh masing-masing variabel bebas secara bersama-sama terhadap variabel terikat seta mencari manakah variabel bebas yang paling berpengaruh terhadap variabel terikat sehingga mendapatkan model tarikan pergerakan kegiatan pelajar. Analisis ini menggunakan analisis regresi linier berganda dengan bantuan Software SPSS Version 17. Analisis regresi linear berganda (Multiple Linear Regression Analysis) yaitu suatu cara yang dimungkinkan untuk melakukan beberapa proses iterasi dengan langkah-langkah sebagai berikut : 1.
Pada langkah awal adalah memilih variabel bebas yang mempunyai korelasi yang besar dengan variabel terikatnya.
2.
Pada langkah berikutnya menyeleksi variabel bebas yang saling berkorelasi, jika ada antara variabel bebas memiliki korelasi besar maka untuk ini dipilih salah satu, dengan kata lain korelasi harus kecil antara sesama variabel bebas.
3.
Pada tahap akhir memasukkan variabel bebas dan variabel terikat ke dalam persamaan model regresi linear berganda yaitu :
Y = a + b1 X1 + b2 X2 + …….. + bn Xn Dimana : Y
= variabel terikat (jumlah kegiatan aktivitas)
a
= konstanta (angka yang akan dicari)
b1,b2,…bn
= koofesien regresi (angka yang akan dicari)
X
= variabel bebas (faktor-faktor berpengaruh)
Faktor-faktor berpengaruh : Y
= Jumlah kegiatan pelajar perhari (orang)
X1
= Jumlah anak di bawah 6 tahun (orang)
X2
= Jumlah anak SD dalam keluarga (orang)
X3
= Jumlah anak SMP dalam keluarga (orang)
X4
= Jumlah anak SMA dalam keluarga (orang)
X5
= Jumlah anak kuliah dalam keluarga (orang)
X6
= Anggota keluarga yang bekerja (orang)
X7
= Jarak tempuh ke sekolah
X8
= Moda yang digunakan
X9
= Waktu berpergian
X10
= Kepemilikan kendaraan 43
Setelah melakukan tahapan diatas dan memperoleh nilai persamaan, maka untuk mengetahui besaran bangkitan pergerakan berdasarkan aktivitas dapat menghasilkan beberapa model bangkitan pergerakan yang berdasarkan aktivitas, dimana terdapat 3 (tiga) macam kategori aktivitas yang akan di tinjau dalam penelitian ini antara lain adalah : a.
Aktivitas mandatory, dimana aktivitas ini adalah aktivitas yang dikerjakan secara rutin dan terus menerus dan memiliki waktu yang terjadwal. Dimana aktivitas ini seperti sekolah.
b.
Aktivitas maitenance, dimana aktivitas ini adalah aktivitas yang tidak rutin dikerjakan dan bersifat tidak terjadwal. Aktivitas ini seperti belanja, les, mengaji, mengantar dan menjemput anggota keluarga.
c.
Aktivitas discretionary, dimana aktivitas ini adalah aktivitas bersenang-senang dan hanya sekali-kali dilaksanakan. Aktivitas ini seperti kegiatan sosial, rekreasi, olahraga, dan aktivitas duduk di warung kopi bersama rekan ataupun keluarga. Kegiatan yang dilakukan oleh pelajar sehari-hari tergolong dalam aktivitas fleksibel. Aktivitas
fleksibel adalah kegiatan yang dilakukan diluar jam sekolah, seperti les atau pergi ke suatu tempat yang tidak dilakukan secara rutin. Adapun hasil yang akan didapat diperoleh dari kuesioner responden yang berisi faktor sosioekonomi dan tata guna lahan yang kemudian dilakukan pengolahan data menggunakan uji korelasi untuk masing-masing variabel sosioekonomi dan tata guna lahan dengan bantuan software SPSS untuk medapatkan koefisien korelasi (R) dan koefisien determinan (R2) dari masing-masing aktivitas sehingga akhirnya diperoleh model bangkitan pergerakan pelajar berdasarkan aktivitas di Kota Banda Aceh. Hasil dari penelitian ini diharapkan dapat menyelesaikan masalah dalam sistem transportasi di Kota Banda Aceh akibat terjadinya tarikan pergerakan saat ini untuk memprediksi karakteristik dan besarnya tarikan pergerakan di masa mendatang.
DAFTAR PUSTAKA Anggraini R, dkk, 2006. A Model of Within-Households Travel Activity Decisions Capturing Interactions between Household Heads, 8th International DDSS Conference, Eindhoven University of Technology. Amelia. A. 2004. Penentuan Model Bangkitan Pergerakan Pada Kawasan Perumahan Di Kota Medan (Studi Kasus : Kawasan Sungal Medan), Tesis Program Magister Manajemen Pembangunan Kota, USU, Medan. Arikunto. S. 2006. Prosedur Penelitian Suatu Pendekatan Praktek, Edisi Revisi VI, Penerbit Rineka Cipta, Jakarta
44
Ashari. H.R. Gumala. 2012. Pemodelan Kebutuhan Pergerakan Berbasis Aktivitas Dari Komplek Perumahan Di Kecamatan Darul Imarah Kabupaten Aceh Besar, Tesis Program Studi Magister Teknik Sipil Unsyiah, Banda Aceh Bruton M.J. 1985, Introduction To Transportation Planning, Hutchinson Technical Education, London. Ikhsani. 2012. Pemodelan Tarikan Pergerakan Di Kota Meureudu Kabupaten Pidie Jaya, Tesis Program Studi Magister Teknik Sipil Unsyiah, Banda Aceh Isya. M. 1998. Model Bangkitan Pergerakan Keluarga Dari Zona Perumahan (Studi Kasus Perumahan Kajhu Aceh Besar), Jurnal Simposium I FSTPT. Desember 1998. Levinson H. S. 1976. Transportation And Traffic Engineering Handbook., New Jersey. LPM ITB. 1997. Modul Pelatihan Manajemen Lalu Lintas Perkotaan,ITB, Bandung. Miro. F. 2005. Perencanaan Transportasi, Penerbit Erlangga, Jakarta. Morlok, E. K. 1991. Pengantar Teknik dan Perencanaan Transportasi, Penerbit Erlangga Jakarta. Patmadjaja H, dkk. 2002. Pemodelan Bangkitan Pergerakan pada Tata Guna Lahan Sekolah Dasar Swasta di Surabaya, DTS. Vol. 4 No. 2. 69-76. September 2002. Sulistyo. J. 2010. 6 Hari Jago SPSS 17. Penerbit Cakrawala. Yogyakarta. Tamin. O.Z. 1997. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Tamin. O.Z. 2000. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Tamin. O.Z. 2008. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Vovsha. P, Petersen. E, Donnely. R. 2004. A Model for Allocation of Maintenance Activities to the Household Members, Papers Presented at the 83th Annual Meeting of the TRB, Washington D.C. Wahyuno. H dan Buchori. I. 1998. Pola Produksi Perjalanan Di Kawasan Permukiman Pinggiran Kota Semarang, Jurnal Simposium I FSTPT, Desember 1998.
45
Article #2 (Draft article to be submitted to National Journal)
Model Bangkitan Aktivitas dari Pekerja dan Non-Pekerja pada Kota Banda Aceh Nanda Susana1, Renni Anggraini2*, M. Isya2 1 Mahasiswa Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh 2 Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh *Email:
[email protected] Abstrak Studi ini memaparkan tentang besarnya tingkat kebutuhan sarana dan prasarana transportasi dapat diketahui dan dihitung dengan mempelajari variasi hubungan antara ciri pergerakannya yang berdasarkan aktivitas penduduk dengan penggunaan tata guna lahan. Dalam hal ini akan dilakukan kajian pergerakan berdasarkan aktivitas penduduk pada kota Banda Aceh. Penelitian ini bertujuan mendapatkan model kebutuhan pergerakan aktivitas dengan memprediksi keputusan yang mengarah ke pola aktivitas penduduk sebagai pekerja dan non pekerja serta mengindentifikasi faktor-faktor dominan yang mempengaharui terjadinya bangkitan aktivitas di kota Banda Aceh. Survei primer dilakukan melalui pengisian kuesioner pada 3200 kepala keluarga (KK) yang bertempat tinggal di kota Banda Aceh. Hasil dari kuesioner ditabulasikan menjadi variabel bebas dan variabel terikat, kemudian dianalisis menggunakan program SPSS-17. Decisions Tree Model digunakan untuk memodelkan bangkitan pergerakan berdasarkan aktivitas pada Kota Banda Aceh. Hasil dari penelitian ini diharapkan dapat mengindentifikasi faktor-faktor dominan yang mempengaruhi pergerakan dari aktivitas sehingga dapat diperoleh tiga kelompok model bangkitan aktivitas bagi aktivitas pergerakan penduduk sebagai pekerja yaitu aktivitas mandatory, aktivitas maintenance, aktivitas discretionary dan dua kelompok model bangkitan aktivitas bagi aktivitas pergerakan penduduk sebagai non-pekerja yaitu aktivitas maintenance dan aktivitas discretionary pada kota Banda Aceh.
Kata kunci : Model bangkitan aktivitas, aktivitas pekerja dan non pekerja.
1.
PENDAHULUAN
46
Pertumbuhan penduduk yang meningkat setiap tahunnya menjadi suatu pemicu perkembangan serta kepadatan pada suatu kota. Perkembangan dan kepadatan suatu kota tidak terlepas dari banyaknya aktivitas penduduk pada suatu zona yang ditimbulkan oleh pemenuhan kebutuhan yang beragam dimana meningkatkan pergerakan pada bagian transportasi. Banyak aktivitas yang dilakukan oleh penduduk baik yang tergolong menjadi rutinitas seperti bekerja maupun yang tidak. Dorongan berbagai aspek juga menjadi salah satu faktor dalam melakukan pergerakan seperti aspek sosial budaya, aspek ekonomi, aspek pemukiman, aspek kependudukan, aspek sarana dan prasarana serta transportasi. Kurangnya fasilitas transportasi yang memadai seiring dengan pertambahan penduduk akan menimbulkan berbagai macam masalah baru dibidang transportasi seperti kemacetan, polusi kendaraan, kebisingan dan lain sebagainya. Pembangkit perjalanan atau bangkitan perjalanan ini berhubungan dengan penentuan jumlah perjalanan keseluruhan yang di bangkitkan oleh suatu kawasan, Miro (2005), Dalam prosesnya, bangkitan perjalanan ini dianalisis secara terpisah menjadi dua bagian yaitu Trip Production dan Trip Attraction. Setiap suatu kegiatan pergerakan mempunyai zona asal dan tujuan, dimana asal merupakan zona yang menghasilkan prilaku pergerakan, sedangkan tujuan adalah zona yang menarik pelaku melakukan kegiatan. Jadi terdapat dua pembangkit pergerakan, yaitu : 1. Trip Production adalah jumlah perjalanan yang dihasilkan suatu zona 2. Trip Attraction adalah jumlah perjalanan yang ditarik oleh suatu zona Trip Production digunakan untuk menyatakan suatu pergerakan berbasis rumah yang mempunyai asal dan/atau tujuan adalah rumah atau pergerakan yang dibangkitkan oleh pergerakan berbasis bukan rumah. Trip Attraction digunakan untuk menyatakan suatu pergerakan berbasis rumah yang mempunyai tempat asal dan/atau tujuan bukan rumah atau pergerakan yang tertarik oleh pergerakan berbasis bukan rumah (Tamin, 1997) Miro (2005), aktivitas meringkas dan menyederhanakan kondisi realistis (nyata) tersebut kita kenal sebagai aktivitas pemodelan. Model merupakan suatu representasi ringkas dari kondisi riil dan berwujud suatu bentuk rancangan yang dapat menjelaskan atau mewakili kondisi riil tersebut untuk suatu tujuan tertentu (Black, 1981). Model merupakan penyederhanaan dari keadaan sebenarnya dan model dapat memberikan petunjuk dalam perencanaan transportasi. Karakteristik sistem transportasi untuk daerah-daerah terpilih seperti CBD (Central Bussiness District) sering dianalisis dengan model. Model memungkinkan untuk mendapatkan penilaian yang cepat terhadap alternatif-alternatif transportasi dalam suatu daerah (Morlok, 1991). Tahapan pemodelan bangkitan pergerakan bertujuan meramalkan jumlah pergerakan pada setiap zona asal dengan menggunakan data rinci mengenai tingkat bangkitan pergerakan, atribut sosio ekonomi, serta tata guna lahan. 47
2.
METODE
2.1
Klasifikasi Aktivitas Didalam penelitian transportasi, di jelaskan bahwa perjalanan berasal dari rumah tangga yang
dilakukan dibawah spasial dan kepentingan sementara. Menurut Anggraini, et all (2006), proses penjadwalan aktivitas terdiri dari 4 (empat) komponen utama: 1.
Aktivitas bekerja (termasuk pemilihan waktu, lamanya, lokasi dan pemilihan moda transportasi untuk masing-masing perjalanan)
2.
Aktivitas tetap sekunder (termasuk pemilihan waktu, lamanya dan lokasi)
3.
Aktivitas fleksibel (termasuk pemilihan waktu, lamanya dan lokasi
4.
Keputusan perubahan perjalanan dan pemilihan moda transportasi untuk masing-masing perjalanan
Kegiatan yang dilakukan pekerja dan non pekerja dibagi ke dalam 2 (dua) kelompok. Aktivitas tetap dan aktivitas fleksibel, aktivitas tetap atau mandatory yaitu adalah kegiatan rutin/ tetap yang dilakukan sehari –hari seperti bekerja oleh individu perorangan atau pekerja. Aktivitas fleksibel atau maintenance yaitu kegiatan seperti belanja harian, belanja yang tidak dilakukan perhari, membawa dan mengantar anak atau orang lain dan sebagainya yang dilakukan oleh non pekerja. Seperti disebutkan dalam Vovsha, dkk (2004), aktivitas maintenance rumah tangga dapat lebih lanjut di bagi dalam tiga kategori : belanja, antar jemput, dan aktivitas maintenance lainnya. Belanja penting untuk memunuhi kebutuhan rumah tangga seperti makanan, pakaian, perlengkapan rumah, dan sebagainya. Antar jemput diistilahkan aktivitas membawa dan mengambil. Untuk mengindentifikasi dan menghitung data aktivitas yang ada maka berbagai macam aktivitas dikelompokkan menjadi 8 aktivitas yang terdapat pada Tabel 2.1 berikut ini : Tabel 2.1. Klasifikasi Aktivitas Individu (I) No
Aktivitas
dan Keluarga
Ruang Lingkup Kegiatan
(K) 1.
Bekerja
I
Full – time dan Part time
2.
Bisnis
I
Kerja – Relasi
3.
Pendidikan
I
Aktivitas rutin (sekolah, les dll)
4.
Antar Jemput
K
Antar jemput anak sekolah dll
5.
Berbelanja
K
Ke Supermarket, Mall, Pasar dll
6.
Kegiatan Sosial
K
Mengunjungi teman, keluarga dll
48
7.
Layanan pengantar
8.
Rekreasi
I I/KK
Layanan delivery Menonton, ke museum, pustaka dan melakukan kegiatan berolahraga dll
Sumber : Anggraini, et. all 2007
2.2.
Metode Pengambilan Data
Pengambilan data bagi suatu studi transportasi pada dasarnya bukan merupakan prosedur yang sembarangan, tetapi merupakan sekumpulan langkah-langkah yang saling terkait satu sama lain dengan hasil final untuk memperoleh data yang diinginkan. Pengambilan data pada penelitian ini dilakukan dengan menggunakan metode survei dengan menyebarkan kuesioner ke rumah tangga. Metode survei yang digunakan metode survei wawancara rumah tangga yang telah dilakukan pada penelitian sebelumnya, dan dilakukan metode survei asal-tujuan untuk menetapkan titik-titik penelitian yang dilakkan pada kawasan-kawasan pemukiman yang sangat potensial untuk menimbulkan perjalanan dengan berbagai aktivitasnya. Oleh karena itu, data yang diperoleh dari survei ini berguna sebagai input data untuk tahap bangkitan perjalanan, karena zona pemukimanlah yang memproduksi perjalanan. Jenis data yang digunakan dalam penelitian ini adalah data primer dan data sekunder. Data primer adalah data yang diperoleh langsung dari responden atau obyek yang diteliti, atau ada hubungannya dengan yang diteliti. Dalam penulisan ini data primer yang dimaksud adalah data yang sumbernya diperoleh langsung dari responden/penghuni perumahan dengan kuesioner, yaitu data jumlah anggota keluarga (orang), jumlah penghasilan keluarga (rupiah), jumlah kepemilikan kendaraan (unit), jumlah keluarga yang bekerja (orang), jumlah keluarga yang sekolah (orang), jenis pekerjaan, umur, jumlah kegiatan dalam satu hari, jarak tempuh perjalanan. Sedangkan data sekunder adalah data yang lebih dulu dikumpulkan dan dilaporkan oleh orang atau instansi diluar diri peneliti sendiri, walaupun yang dikumpulkan itu sesungguhnya data yang asli. Data sekunder diperoleh dari instansi-instansi terkait dan perpustakaan. Di dalam penelitian ini data sekunder sumbernya lebih banyak diperoleh dari Kantor Badan Pusat Statistik Kota Banda Aceh. Pada penelitian ini data jumlah kepala keluarga pada Kecamatan yang ditinjau diperoleh dari data penduduk pada tahun 2011.
2.3
Metode Analisis Data
49
Dalam pemodelan bangkitan pergerakan, kami menggunakan metode CHAID dimana secara keseluruhan bekerja untuk menduga sebuah variabel tunggal, disebut sebagai variabel dependen, yang didasarkan pada sejumlah variabel-variabel yang lain, disebut sebagai variabel-variabel independen. CHAID merupakan suatu teknik iteratif yang menguji satu-persatu variabel independen yang digunakan dalam klasifikasi, dan menyusunnya berdasarkan pada tingkat signifikansi statistik chi-square terhadap variabel dependennya (Gallagher, 2000). CHAID digunakan untuk membentuk segmentasi yang membagi sebuah sampel menjadi dua atau lebih kelompok yang berbeda berdasarkan sebuah kriteria tertentu. Hal ini kemudian diteruskan dengan membagi kelompok-kelompok tersebut menjadi kelompok yang lebih kecil berdasarkan variabel-variabel independen yang lain. Prosesnya berlanjut sampai tidak ditemukan lagi variabel independen yang signifikan secara statistik. Segmen-segmen yang dihasilkan akan bersifat saling lepas yang secara statistik akan memenuhi kriteria pokok segmentasi dasar (Bagozzi, 1994). Hasilnya juga akan memberikan peringkat pada variabel yang merupakan variabel independen paling signifikan sampai yang tidak signifikan. Dalam analisis CHAID variabel yang digunakan dibedakan atas variabel terikat (variabel dependen) dan variabel bebas (variabel independen). Klasifikasi dalam CHAID dilakukan berdasarkan pada hubungan yang ada antara kedua variabel tersebut, oleh karena itu CHAID termasuk dalam metode dependensi dalam menentukan segmentasi. Menurut Gallagher (2000), CHAID akan membedakan variabel variabel independennya menjadi tiga bentuk yang berbeda, yaitu: 1.
Monotonik : kategori-kategori pada variabel ini dapat dikombinasikan atau digabungkan oleh CHAID hanya jika keduanya berdekatan satu sama lain, yaitu variabel-variabel yang kategori nya mengikuti urutan aslinya (data ordinal), contohnya: usia atau pendapatan.
2.
Bebas : kategori-kategori pada variabel ini dapat dikombinasikan atau digabungkan walaupun keduanya berdekatan atau tidak satu sama lain (data nominal), contohnya: pekerjaan, kelompok etnik, dan area geografis
3.
Mengambang (floating): kategori-kategori pada variabel ini akan diperlakukan seperti monotonik kecuali untuk kategori yang terakhir (yaitu missing value), yang dapat berkombinasi dengan kategori manapun. Hasil dari analisis CHAID akan ditampilkan dalam diagram pohon. Dimana diagram ini sering
digunakan untuk permodelan pergerakan berbasis aktivitas. Untuk pengklasifikasian aktivitas juga dapat menggunakan metode decision tree yang paling populer digunakan karena pembangunannya relatif cepat, hasil dari model yang dibangun mudah dipahami.
50
Decision tree merupakan salah satu metode klasifikasi yang menggunakan representasi struktur pohon (tree) dimana setiap node mempresentasikan atribut, cabangnya memepresentasikan nilai dari atribut, dan daun mempresentasikan kelas. Node yang paling atas dari decision tree disebut sebagai root. Pada decision tree terdapat 3 jenis node, yaitu; a) Root Node yaitu merupakan node paling atas, pada node ini tidak ada input dan bisa tidak mempunyai output atau mempunyai output lebih dari satu. b) Internal Node yaitu merupakan node percabangan, pada node ini hanya terdapat satu input dan mempunyai output minimal dua. c) Leaf node atau terminal node merupakan node akhir, pada node ini hanya terdapat satu input dan tidak mempunyai output.
3.
HASIL
Keseluruhan pengambilan sampel pada Kota Banda Aceh dapat dilihat pada Tabel 3.1 yaitu:
Tabel 3.1. Data Jumlah Pengambilan Sampel Kota Banda Aceh No.
Kecamatan
1 2 3 4 5 6 7 8 9
Meuraxa Jaya Baru Banda Raya Baiturrahman Lueng Bata Kuta Alam Kuta Raja Syiah Kuala Ulee Kareng JUMLAH
Jumlah Populasi 16.861 22.535 21.369 31.073 24.132 43.184 10.672 35.648 23.088 228.562
Jumlah Sampel Jumlah Sampel Keterangan Responden KK 1.356,70 339,17 1.384,73 346,18 1.380,11 345,03 1.408,50 352,12 1.390,38 347,60 Banda Aceh 1.426,62 356,66 1.296,25 324,06 1.416,73 354,18 1.386,77 346,69 12.446,81 3.111,70
Dalam penelitian ini, jumlah sampel yang didapat dari populasi Kota Banda Aceh yaitu sebanyak 3111,70 KK dengan mengasumsikan jumlah dari setiap 4 responden = 1 kepala keluarga (KK) sehingga sampel dibulatkan menjadi 3200 kepala keluarga (KK). Hasil yang akan didapat diperoleh dari kuesioner responden yang berisi faktor sosioekonomi dan tata guna lahan yang kemudian dilakukan pengolahan data menggunakan uji korelasi untuk masingmasing variabel sosio ekonomi dan tata guna lahan dengan bantuan software SPSS untuk mendapatkan jumlah alternative dari keputusan masing-masing Aktivitas dari pekerja dan non pekerja sehingga akhirnya diperoleh 3 kelompok model bangkitan aktivitas bagi aktivitas pergerakan penduduk sebagai
51
pekerja yaitu aktivitas mandatory, aktivitas maintenance, aktivitas discretionary dan 2 model bangkitan aktivitas bagi aktivitas pergerkan penduduk sebagai non pekerja yaitu aktivitas maintenance dan aktivitas discretionary pada kota Banda aceh yang mana total keseluruhan menjadi 5 model bangkitan pergerakan.
DAFTAR PUSTAKA Anggraini R, dkk, 2006. A Model of Within-Households Travel Activity Decisions Capturing Interactions between Household Heads, 8th International DDSS Conference, Eindhoven University of Technology. Anggraini R, dkk, 2007. Modeling Household Activity Generation and Allocation Decisions Using Decision Tree Induction, Paper Presented in WCTR Conference 2007, Berkeley USA. Bhat, C. R. and R. Misra (2000). "Nonworker activity-travel patterns: Organization of activities”, presented at the 79th Annual Meeting of the Transporation Research Board, Washington, D.C., January. Bruton M.J. 1985, Introduction To Transportation Planning, Hutchinson Technical Education, London. Kitamura, R. (1995). “Applications of models of activity behavior for activity based demand forecasting,” presented at the Activity-Based Travel Forecasting Conference, New Orleans, Louisiana Miro. F. 2005. Perencanaan Transportasi, Penerbit Erlangga, Jakarta. Morlok, E. K. 1991. Pengantar Teknik dan Perencanaan Transportasi, Penerbit Erlangga Jakarta. Morlok, E. K. 1995. Pengantar Teknik dan Perencanaan Transportasi, Penerbit Erlangga Jakarta. Tamin. O.Z. 1997. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Tamin. O.Z. 2008. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Vovsha. P, Petersen. E, Donnely. R. 2004. A Model for Allocation of Maintenance Activities to the Household Members, Papers Presented at the 83th Annual Meeting of the TRB, Washington D.C.
52
Article #3 (Draft article to be submitted to National Journal)
Analisis Keputusan Pengalokasian Kendaraan Pada Rumah Tangga yang Kekurangan Kendaraan (Vehicle Allocation Decisions in Vehicle Deficient Household) Muti Handayani1, Renni Anggraini2*, Sofyan M.Saleh2 1 Mahasiswa Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh 2 Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh *Email:
[email protected]
Abstrak Peningkatan penduduk cenderung akan meningkatkan kebutuhan akan pergerakan seiring dengan munculnya banyak perumahan-perumahan yang akan menambah jumlah pergerakan dari aktivitas yang dilakukan sehari-hari. Kebutuhan akan pergerakan ini juga disertai dengan bertambahnya jumlah kendaraan dari setiap rumah atau keluarga baik kendaraan roda 4 (empat) maupun kendaraan roda 2 (dua) didalam rumah tangga. Penelitian ini dilakukan bertujuan untuk mendapatkan keputusan siapa yang akan mendapatkan kendaraan dalam setiap rumah tangga berdasarkan aktivitas yang dilakukan oleh penghuni perumahan pada waktu yang sama dengan tujuan dan lokasi yang berbeda dan survey ini dilakukan melalui penyebaran kuisioner pada keluarga yang bertempat tinggal di Kecamatan-kecamatan di kota Banda Aceh. Tingkat aktivitas di rumah tangga ditunjukkan pada pengaruh pengalokasian kendaraan. Baik lelaki dan perempuan melakukan aktivitas yang sama setiap harinya, maka dari itu pengalokasian kendaraan didalam rumah tangga pun meningkat seiring banyaknya pergerakan yang terjadi, kemungkinan lelaki dan perempuan memperoleh peningkatan dengan meningkatnya waktu perjalanan secara monoton, sosial ekonomi, dan juga faktor situasi berpengaruh terhadap keputusan pengalokasian kendaraan.
Kata kunci : Model bangkitan pergerakan,pengalokasian kendaraan
1.
Pendahuluan Keberhasilan suatu wilayah ditandai dengan meningkatnya pertumbuhan ekonomi yang berarti juga
peningkatan standar kehidupan masyarakat. Hal ini dapat dilihat dari munculnya pusat-pusat kegiatan, yang berarti menambah intensitas pergerakan dan aktivitas baik orang maupun barang ke pusat kegiatan atau sebaliknya. Dengan berkembang dan meningkatnya suatu kawasan di Kota Banda Aceh dan sebagian Kecamatan, maka semakin meningkat jumlah pengalokasian kendaraan di dalam setiap rumah tangga, maka dari itu suatu wilayah akan menimbulkan beragam aktivitas dan juga yang tidak disertai dengan penambahan fasilitas yang memadai dan ini akan menimbulkan masalah. Peningkatan penduduk cenderung akan meningkatkan kebutuhan akan pergerakan seiring dengan munculnya banyak perumahan-perumahan yang akan menambah jumlah pergerakan dari aktivitas yang
53
dilakukan sehari-hari. Kebutuhan akan pergerakan ini juga disertai dengan bertambahnya jumlah kendaraan dari setiap rumah atau keluarga baik kendaraan roda 4 (empat) maupun kendaraan roda 2 (dua) didalam rumah tangga. Penelitian ini dilakukan bertujuan untuk mendapatkan keputusan siapa yang akan mendapatkan kendaraan dalam setiap rumah tangga berdasarkan aktivitas yang dilakukan oleh penghuni perumahan melalui penyebaran kuisioner pada keluarga yang bertempat tinggal di Kecamatan di kota Banda Aceh antara lain Kecamatan Kuta Alam, Syiah Kuala, Baiturrahman dan Kecamatan lainnya. Keputusan aktivitas didalam setiap rumah tangga saling berhubungan dan juga jumlah kepemilikan kendaraan, baik yang memiliki SIM A (Surat Izin Mengemudi) untuk kendaraan roda 4 (empat) ataupun SIM C kendaraan roda 2 (dua). Tingkat aktivitas di rumah tangga ditunjukkan pada pengaruh pengalokasian kendaraan. Baik lelaki dan perempuan melakukan aktivitas yang sama setiap harinya, maka dari itu pengalokasian kendaraan didalam rumah tangga pun meningkat seiring banyaknya pergerakan yang terjadi, kemungkinan lelaki dan perempuan memperoleh peningkatan dengan meningkatnya waktu perjalanan secara monoton, sosial ekonomi, dan juga faktor situasi berpengaruh terhadap keputusan pengalokasian kendaraan. Banyaknya lalu lintas yang ditimbulkan oleh suatu zona atau daerah per satuan waktu. Jumlah lalu lintas bergantung pada kegiatan
kota, karena penyebab lalu lintas ialah kebutuhan manusia untuk
melakukan kegiatan berhubungan dengan mengangkut barang (Warpani, 1990: 17). Tujuan dasar tahap bangkitan pergerakan adalah menghasilkan model hubungan yang mengaitkan parameter tata guna lahan dengan jumlah pergerakan yang menuju ke suatu zona atau jumlah pergerakan yang meninggalkan suatu zona (Tamin O.Z, 2000: 111).
2.
Metode penelitian
2.1
Klasifikasi Aktivitas Didalam
penelitian
transportasi,
di
jelaskan
bahwa
perjalanan
berasal dari rumah tangga yang dilakukan dibawah spasial dan kepentingan sementara. Menurut Anggraini, dkk (2006), proses penjadwalan aktivitas terdiri dari 4 (empat) komponen utama: 1.
Aktivitas bekerja (termasuk pemilihan waktu, lamanya, lokasi dan pemilihan moda transportasi untuk masing-masing perjalanan)
2.
Aktivitas tetap sekunder (termasuk pemilihan waktu, lamanya dan lokasi)
3.
Aktivitas fleksibel (termasuk pemilihan waktu, lamanya dan lokasi)
4.
Keputusan perubahan perjalanan dan pemilihan moda transportasi untuk masing-masing perjalanan
54
Pada penelitian ini ditinjau kegiatan yang dilakukan pekerja dan non pekerja yaitu aktivitas yang dibagi ke dalam 2 (dua) kelompok. Aktivitas tetap dan aktivitas fleksibel, aktivitas tetap atau mandatory yaitu adalah kegiatan rutin/ tetap yang dilakukan sehari –hari seperti bekerja oleh individu perorangan atau pekerja. Aktivitas fleksibel atau maintenance yaitu kegiatan seperti belanja harian, belanja yang tidak dilakukan perhari, membawa dan mengantar anak atau orang lain dan sebagainya yang dilakukan oleh non pekerja.
2.2
Pengalokasian Kendaraan (Vehicle Allocation) Di Indonesia gender sepertinya masih berperan, sebuah analisis deksriptif menunjukkan bahwa
laki-laki lebih sering mendapatkan kendaraan untuk beraktifitas dibandingkan dengan perempuan di dalam rumah tangga. Keputusan pengalokasian kendaraan ini dianggap sebagai unsur yang lebih meliputi proses penjadwalan kegiatan. Sejumlah besar faktor yang berpotensi mempengaruhi keputusan alokasi mobil di rumah tangga yang harus dipertimbangkan. Ini faktor berhubungan dengan aktivitas-jadwal, pengaturan ruang-waktu, dan individu dan rumah tangga yang berkarakteristik. Untuk mendapatkan pemahaman yang lebih baik bagaimana keputusan dibuat dalam rumah tangga, kita fokus pada rumah tangga yang terdiri dari dua (pria-wanita) kepala rumah tangga. Keduanya adalah pengemudi, dan rumah tangga memiliki satu mobil (Anggraini, dkk, 2009) Menurut Anggraini, dkk (2009), keputusan pengalokasian kendaraan berfokus pada rumah tangga yang kekurangan kendaraan (yaitu pengemudi melebihi jumlah mobil) dan melibatkan keputusan bersama antara laki-laki dan perempuan. Seperti
yang ditunjukkan pada penelitian sebelumnya di
Belanda oleh Anggraini, dkk (2009) total sampel diekstraksi dari data yang diambil termasuk 28.600 rumah tangga. Mengingat tujuan penelitian ini, di beberapa rumah tangga sebagai berikut : (1) ada dua kepala keluarga dalam rumah tangga, (2) ada satu mobil di rumah tangga; (3) kedua kepala adalah pengemudi dan (4) kedua kepala keluarga memiliki kegiatan/ aktivitas di luar pekerjaan di hari yang sama.
55
Non-work tour Male
Non-work tour
Work tour
Case #1
Case #2
Case #3
Gambar 2.6. Contoh jadwal keputusan pengalokasian kendaraan di dalam rumah tangga (Sumber : Anggraini dkk, 2009)
Seperti yang muncul, 3.190 rumah dan 4.049 jumlah kasus sesuai dengan kriteria tersebut. Model ini mencakup pilihan dimana ada 156 kepala rumah tangga yang tidak menggunakan mobil, tetapi beberapa alat transportasi lain sebagai gantinya. Oleh karena itu, pilihan keputusan adalah laki-laki, perempuan, dan keduanya menggunakan kendaraan lain (contohnya : jalan kaki, sepeda, angkutan umum dan lain-lain). Dalam penggunaan kendaraan didalam rumah tangga dapat dilihat pengalokasiannya antara lakilaki dan wanita, mereka memiliki jam yang sama setiap harinya dari jam 8.00 pagi sampai jam 14.00 siang tetapi wanita memiliki waktu diluar kegiatan bekerja di jam 11.00 sampai jam 14.00 dan jam 16.00 sampai 18.00 sehingga pengalokasian kendaraan akan terbagi dengan laki-laki yang sedang bekerja. Sedangkan laki-laki memiliki waktu diluar kegiatan bekerja antara jam 17.00 sore dan jam 20.00 malam. Disini didalam pengalokasian kendaraan kita melihat kepemilikan SIM (Surat Izin Mengemudi) yang sudah berhak memiliki atau sudah cukup umur untuk mendapatkan SIM. Di Indonesia, Surat Izin Mengemudi (SIM) adalah bukti registrasi dan identifikasi yang diberikan oleh Polri kepada seseorang yang telah memenuhi persyaratan administrasi, sehat jasmani dan rohani, memahami peraturan lalu lintas dan terampil mengemudikan kendaraan bermotor. Setiap orang yang mengemudikan Kendaraan Bermotor di Jalan wajib memiliki Surat Izin Mengemudi sesuai dengan jenis Kendaraan Bermotor yang dikemudikan. 56
Persyaratan permohonan SIM perseorangan di Indonesia berdasarkan Pasal 81 ayat (2), (3), (4), dan (5) UU No. 22 Tahun 2009 adalah usia 17 Tahun untuk SIM A, C dan D. Tapi beda hal nya dengan persyaratan permohonan bagi SIM umum berdasarkan Pasal 83 ayat (1), (2), dan (3) UU No. 22 Tahun 2009 adalah usia 20 tahun untuk SIM A.
2.3
Analisa Data Untuk menjawab perumusan masalah yang telah ditetapkan, yaitu berapa besar pengaruh variabel
mengenai bangkitan pergerakan (X) seperti : jumlah penghasilan keluarga (rupiah), jumlah kepemilikan mobil (unit), jumlah kepemilikan sepeda motor (unit), jumlah kepemilikan SIM A dan C, jenis pekerjaaan, umur, jarak tempuh ke tempat aktivitas, jumlah kegiatan berdasarkan aktivitas dalam satu hari (Y), perlu dilakukan beberapa tahapan penting untuk menganalisis data yang diperoleh melalui survei kuesioner. Uji korelasi dan proses kalibrasi dilakukan dengan menggunakan bantuan software SPSS (Statistical Product and Service Solution) yaitu suatu program statistik yang mampu memproses data statistik secara tepat dan tepat serta menyajikannya dalam berbagai output yang dikehendaki para pengambil keputusan (Joko Sulistyo, 2010) dan diikuti dengan metode Decision Tree (CHAID) Pengolahan Data dan Analisis, ada beberapa tahap analisis yang perlu dilakukan yaitu : a. Tahap pertama adalah analisis bivariat, yaitu analisis data dua variable bertujuan untuk mencari keterkaitan hubungan antara variable bebas dan variable terikat untuk masing-masing data variable, analisis ini menggunakan Chi Square atau metode Chaid dengan bantuan program SPSS. b. Tahap Kedua adalah analisis multivariate, yaitu untuk data lebih atau dua variable yang bertujuan untuk mencari pengaruh masing-masing variable terikat serta mencari manakah variable bebas yang paling berpengaruh terhadap variable terikat sehingga mendapatkan model kepemilikan mobil dan motor. c. Tahap ketiga adalah anilisis Peramalan, yaitu analisi yang dilakukan untuk meramalkan kepemilikan mobil dan sepeda motor lima tahun yang akan dating dengan menggunakan persamaan model hasil dari analisa multivariate. Variabel
pendapatan rumah tangga
diasumsikan terjadi peningkatan untuk lima tahun kedepan, sedangkan untuk variabel lainnya tetap menggunakan variabel hasil survey yang telah dilakukan.
3.
Hasil dan Pembahasan Adapun hasil yang akan didapat diperoleh dari kuesioner responden yang berisi faktor sosio
ekonomi dan pengalokasian kendaraan yang kemudian dilakukan pengolahan data menggunakan uji korelasi untuk masing-masing variabel sosio ekonomi dengan bantuan software SPSS dengan metode 57
Decision Tree yang dianalisi melalui CHAID untuk medapatkan tingkat pengaruh dari beberapa faktor yang diteliti terhadap bangkitan aktivitas dan pengalokasian kendaraan didalam rumah tangga yang dilakukan di
wilayah kota Banda Aceh Nantinya kita akan mendapatkan pengaruh pengalokasian
kendaraan terdiri dari tiga yaitu : 1. Satu kendaraan roda empat dengan dua pengemudi. 2. Satu kendaraan roda dua dengan dua pengemudi. 3. Satu kendaraan rodan empat dan satu kendaraan roda dua serta dua pengemudi.
DAFTAR PUSTAKA Anggraini R, dkk, 2006. A Model of Within-Households Travel Activity Decisions Capturing Interactions between Household Heads, 8th International DDSS Conference, Eindhoven University of Technology. Anggraini R, dkk, 2009. Car Allocation Decisions In Car Deficient Household non- work tour, International DDSS Conference, Eindhoven University of Technology. Arikunto. S. 2006. Prosedur Penelitian Suatu Pendekatan Praktek, Edisi Revisi VI, Penerbit Rineka Cipta, Jakarta Arentze T.A dan Timmernans, 2006. Transport Stated Choice Responses, London Bruton M.J. 1985, Introduction To Transportation Planning, Hutchinson Technical Education, London. Bagozzi, R.P. 1994, Research Oxford Advanced Methods Of Marketing, London. Sulistyo. J. 2010. 6 Hari Jago SPSS 17. Penerbit Cakrawala. Yogyakarta. Lehman. T dan Enheler. D. 2001, Responden Profilling with CHAID and dependency, New York Levinson H. S. 1976. Transportation And Traffic Engineering Handbook., New Jersey. LPM ITB. 1997. Modul Pelatihan Manajemen Lalu Lintas Perkotaan, ITB,
Bandung
Miro. F. 2005. Perencanaan Transportasi, Penerbit Erlangga, Jakarta. Morlok, E. K. 1991. Pengantar Teknik dan Perencanaan Transportasi, Penerbit Erlangga Jakarta. Myers. J . G 1996, Principle Of Corporations Finance, New York Tamin. O.Z. 2008. Perencanaan dan Pemodelan Transportasi, Penerbit ITB, Bandung. Vovsha. P, Petersen. E, Donnely. R. 2004. A Model for Allocation of Maintenance Activities to the Household Members, Papers Presented at the 83th Annual Meeting of the TRB, Washington D.C.
58
Article #4 (Draft article to be submitted to International Conference)
Individual and Household Travel Survey (Case Study Banda Aceh City) 1
Renni Anggraini, 2H.J.P Timmermans 1 Department of Civil Engineering, Faculty of Engineering, Syiah Kuala University, Banda Aceh 2 Faculty of Architecture, Building & Planning, Eindhoven University of Technology, The Netherlands Corresponding Author:
[email protected]
Abstract The travel survey of person and household will be conducted in Banda Aceh City and surrounding city. The survey is about collecting information regarding person or household day-to-day travel; how they travel, where, when, with whom and so on. The results will provide a representation of the actual travel patterns of all types of people. The given information will include about various ways of people using road network, so that the policy makers will be able to use it for city planning and design work to help reduce traffic congestion and prevent road accidents including developing safe cycling and pedestrian networks, developing road safety policy, and improvement of public transport. This study reported the methodology, sampling technique and questionnaire design for travel survey in Banda Aceh City. It is expected that the study will provide a better way for survey procedures and become the first travel survey of person and household in Banda Aceh City. Keywords: individual, household, activity-travel diary survey, Banda Aceh City
Introduction The current generation of transportation forecasting models is based on activities and considers travel as a derivative of the need to conduct activities at different locations. In addition to predicting transport demand, the approach used is a key stepping stone for long terms goals, to modeling the impact of global climate change, contribution of transport to environmental issues and social sustainability. The research project aims to develop an operational system of activitybased travel demand modeling (Albatross) based on local research and local data, in particular for Banda Aceh city. The method uses a CHAID decision tree induction method. It is expected that the model will also be a benchmark and applicable to use in other cities in Indonesia.
Descriptions of Banda Aceh City Banda Aceh was hardly known in international discussion until 26 December 2004, the day when the Indian Ocean Earthquake struck off the western coast of Sumatra. Banda Aceh was the closest major city to the earthquake's epicenter and suffered further damage when a tsunami 59
struck shortly afterward. It was the worst hit area out of all the locations hit. More than 167,000 people died and many more were injured. Banda Aceh is the capital city of Province of Aceh and the largest city in Aceh, located on The Island of Sumatra, Indonesia. The city was originally named Kuta Raja, which means "City of the King", in reference to the founding of the Aceh Sultanate. Later, its name was changed to Banda Aceh, with the first part of the name coming from the Persian bandar meaning "port" or "haven". It is also proudly referred to as "Port to Mecca", as Islam first arrived in Aceh before spreading throughout Southeast Asia.
Figure 1. Location of Banda Aceh (Province of Aceh) in Map of Indonesia Source: Dinas Bina Marga of Aceh (2013)
The population was approximately 228.562 (BPS, 2012). Banda Aceh is located at the northwestern tip of Indonesia at the mouth of the River Krueng Aceh. The provincial capital of Banda Aceh is clearly the largest and most lively city in the region. The first part of its name comes from the Persian bandar and means "port" or "haven". The city has a mix of old and 60
grandiose architecture from the golden days and more modern buildings. The maybe most famous landmark here is the Mesjid Raya Baiturrahman mosque in the southern part of the city, behind the mosque is a large market. Most of the hotels and restaurants are located in the northern part. North of Banda Aceh is the beautiful Pulau Weh island, many come to Aceh just to visit this island with it's lovely beaches and good diving conditions.
Survey Method Survey method is carried out by doing home interview survey to residents of Banda Aceh city. It covers 9 sub districts (kecamatan) in Banda Aceh city, as follows: 1. Kecamatan Meuraxa 2. Kecamatan Jaya Baru 3. Kecamatan Banda Raya 4. Kecamatan Baiturrahman 5. Kecamatan Lueng Bata 6. Kecamatan Kuta Alam 7. Kecamatan Kuta Raja 8. Kecamatan Syiah Kuala 9. Kecamatan Ulee Kareng Home interview survey containing questionnaires will be performed to densely residential areas in those kecamatans, that is potential to generate/produce trip along with its activities. The questionnaire is dispatched to those households selected as samples. The survey is conducted on a regular basis to obtain travel and activity information of residents in Banda Aceh city. It is a household survey where data is collected of all household members for the diary day as well as general information about household and individual attributes such as, gender, age, vehicle ownership and driving license ownership, home location, individual income, occupation, number of working hours per week, etc. Respondents were also requested to give information about all trips made on a designated day as well as on the activities conducted on trip destinations. Information for each trip includes start time, trip purpose, destination, activity type at the destination, and transport mode. Situational variables are reported as well. All in all, this survey provides a comprehensive data source to analyze activity-travel behavior of Banda Aceh city residents. 61
Sample Size According to Stopher, et al (2008), sample size is probably the most controversial item in household travel surveys and one on which there is virtually no agreement, as evidenced by samples ranging from a few hundred to as much as 20,000 households. Procedures to develop suitable sample sizes are important issue, not only in the transportation field but also in the survey sampling literature. Obviously, there is such a wide variation in chosen sample sizes for household travel surveys arises from at least two issues: (1) available budget and (2) political rather than statistical justification of a particular sample size. The more data available, the better results will be acquired. The population of Banda Aceh city in 2011 can be seen in Table 1.
Table.1. Population of Banda Aceh City in Year 2011 No. 1. 2. 3. 4. 5. 6. 7. 8. 9.
Kecamatan Total Population (person) Meuraxa 16.861 Jaya Baru 22.535 Banda Raya 21.369 Baiturrahman 31.073 Lueng Bata 24.132 Kuta Alam 43.184 Kuta Raja 10.672 Syiah Kuala 35.648 Ulee Kareng 23.088 Jumlah Total Populasi 228.562 Source: Central Bureau of Statistics (2012) To determine the size of sample in a study, it uses a formula based on the proportion examined by Isaac & Michael. (Arikunto, 2006). Systematically, the amount of sample can be formulated as follows: S=
λ2 N P (1− P) d 2 ( N −1) + λ2 P (1− P)
where: S = number of samples; N = number of population; λ2 = chi-square; d = degree of accuracy, can be 1%, 5% and 10%; P = proportion of the population;
62
Considering the population of Banda Aceh City and available budget from this research project, we decide to take sample of about 3,200 households for the first year project. It will be added more households for a following year. The population of Banda Aceh city in 2011 can be seen in Table 1. Activity-based models differ considerably from the traditional transportation forecasting methodologies in the sense that these models attempt to predict the interdependencies between the many facets of activity profiles. The difference with previous models is that the timing and scheduling of activities constitute the core of activity-based models. Consequently, the data required to estimate any activity-based model differ from the data necessary to calibrate conventional models. In particular, data on activity patterns are required in order to estimate and validate an activity-based model of transportation demand. The specific data requirements depend on the theoretical principles embedded in the model. As it has been noted, the objective of a full activity-based model is to predict which activities are conducted where, when, for how long, with whom, and the involved of transport mode, and thus requires data on all these facets. Hence, as in common practice, it needs to collect activity diary data form household and individual to estimate activity-based models. In addition to the activity diary data, the estimation of the model requires space-time environment data. We use this term to denote the physical system, the transportation system, and the institutional context. The physical system consists of the set of locations, where particular activity can be conducted. The transportation system requires information about public transport attributes including route, travel times and costs. The institutional context reflects to the opening hours, and hence it will be highly influential to the space-time prism that individual and household are facing. For those reasons, the space-time environment data would be incorporated in the model, either as independent variables or as constraints, and it means that data about these factors need to be collected. Activity diary data and space-time environment data will be possibly collected at the same time. Having prepared the questionnaire format and defined the sample among the population available, the questionnaire will be distributed to Banda Aceh city residents. By the time of collecting the activity-diary data, the space-time environment data is also collected.
63
Article #5 (Draft article to be submitted to International Journal)
Modeling Within-Household Joint Activity Participation Dessy Amalia1, Renni Anggraini2*, Sofyan M.Saleh2 1 Mahasiswa Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh 2 Program Studi Magister Teknik Sipil Universitas Syiah Kuala Jl. Tgk Syech Abdur Rauf no. 7, Darussalam, Banda Aceh *Email:
[email protected] Abstract This paper describes an empirical derivation of activity participation at the household level taking into account joint activity of household members in discretionary activities. The underlying activity-based approach allows for more detailed assessment made on the field of transportation policy, especially in the context of household-level decision making and to determine the level of transportation infrastructure required. The aims of this study is to develop a model of activity participation and to identify the factors that influence the movement and get the level of influence of these factors on seizure activity undertaken jointly between family members by type of household. In this research, observation of the movement of individuals is done by using questionnaires and interviews at the 3200 house where located on the nine districts in the city of Banda Aceh. The questionnaire is grouped by activities conducted jointly and household type of a different amount of activity. Results from the questionnaires were tabulated into independent variables and the dependent variable, and then analyzed by the method of decision tree induced by CHAID method using SPSS - 17. The research results will be obtained five types of seizure models activities carried out with family members for non - work purposes in the city of Banda Aceh for each - each type of household , the household without children , households with less than two children and households having more than two children . Types of households with children divided according to the level of the child's age households with children under 12 years old and who have children over 12 years. Key words: joint activity participation, household decision making, activity-based approach, Decision Tree, CHAID
1.
Introduction The rapid development of technology is triggering the development of a city . In line with the
development of a city , an increase in population will also increase the impact on travel needs . Activity further away from home will improve the long journey, it leads to problems of transportation, such as congestion, pollution from vehicles, noise and others. In many cases, individuals who are influential not only in making decisions but the relationship between family members. Because in general, in addition to work activities carried out jointly between family members, such as social activities, recreation, and so forth. Furthermore, the level of the child's age in the household also affects the activity of which is done 64
with, because the adult child is then likely do activities with their parents will be smaller than the younger. While the vast majority of planning organizations continue to rely on traditional models, academic research suggests that activity-based approaches promise greater predictive capability, more accurate forecasts, and especially more realistic sensitivity to
policy changes (Anggraini, 2009).
The model is a simplification of the actual situation and the model can provide guidance in the planning of transport . Characteristics of the transport system for selected areas such as the CBD ( Central Bussiness District ) is often analyzed by the model . Models possible to get a quick assessment of the transportation alternatives in a region ( Morlok , 1995 ) . Recognition of the various interdependencies in activity timing and other travel attributes allow greater realism in models of travel demand. Moreover, activity-based modeling is better suited to current transportation planning interests. In general, activity-based models focus on activities as the unit of analysis as opposed to trips as the unit of analysis in trip-based models. This shift has enabled the models to address issues related to substitution of non-travel alternatives. Focusing on activity episodes also permits the incorporation of constraints such as time constraints related to opening hours, work schedules, expected activity duration, and multi-day scheduling of activities. This approach reflects the complex interaction between activity and travel behavior. Conceptual appeal of this approach comes from the realization that need and desire to participate in activities that are more basic than based on some trips may require participation. By replacing the main emphasis on participation in activities and focus on the sequence or pattern of activity behavior ( using the whole day or longer periods of time as the unit of analysis ) ( Mc.Nally , et . Al , 2007) . Study area is a geographic area in which is located all the zones of origin and destination zones are calculated in the model transportation needs . The most important criteria for the study area is the area that contains the internal zone and the road was significantly affected by the movement of traffic . ( Tamin , 2008) . The aim of introducing more interdependencies in the models is not only concerned with interdependencies in choice facets, but also with interdependencies between the decisions of individuals. It was realized that in many cases, it is not the individual, but rather the household that makes decisions. Households are relevant in at least three situations. First, the activity-travel patterns of household members need to be synchronized in time and space for joint activities. Second, resources may need to be allocated to individual household members. In turn, resource allocation decisions may limit subsequent choices of individual member. Thirdly, some activities are household activities, implying that only one household member has to conduct that activity. In turn, such task allocation decisions influence other aspects of activity-travel programs. 65
2.
METHOD
2.1
Classification of Activities In the transportation studies, described that trip came from households conducted under spatially
and temporary interested. According to Anggraini, et.al (2006), activity scheduling process consists of four (4 ) main components : 1. Work activities ( including timing , duration , location and selection of transportation modes for each trip ) 2. Secondary activity remains ( including the timing , duration and location ) 3. Flexible activities ( including the timing , duration and location ) 4. Decisions and drive change in the selection of transportation modes for each trip As mentioned in Vovsha, et.al (2004), household maintenance activities can be further divided into three categories : shopping , transport, and other maintenance activities. Important for the household such as food, clothing, home furnishings, and so on. To drop off and pick up their school or daycare. Then other maintenance activities such as go to the bank to be grouped together with other activities into a single activity. The first step is to classify the activities undertaken jointly both maintenance and discretionary activities. Activities to be observed are as follows:
Table 1 Activity Classification Household
No
Activity
1
Shop-n-store
HH
2
Service-related
HH
3
Social-joint
HH
4
Leisure-joint
HH
Touring-joint
HH
5
2.2
(HH) Level
Scope of Activities Shopping n stores Renting
movie,
getting
(fast)
food,
institutional purposes (bank, post office, etc) Visiting friends, relatives , etc Sports, café/bar, eating out, movie, museum, library Making a tour by car, bike, or foot (e.g., letting out the dog, etc)
Method of Data Collection
Retrieval of data for a transportation study is essentially not an arbitrary procedure, but it is a set of measures that are interrelate to each other with the final results to obtain the desired data .
66
Data used in this study is primary data and secondary data. Primary data is data obtained directly from respondents or objects under study , or anything to do with the investigation. In this study, the primary data is the data source is obtained directly from the respondents / residents of the housing with the questionnaire , the data on number of family members (people), the amount of family income (Rupiah), number of children (people), the level of the child's age (age), type of work, amount of work being done with in a day, trip mileage . While secondary data is data first collected and reported by the person or agency. In this study secondary data obtained from the relevant agencies and libraries. In this study secondary data obtained from many sources over the Office National Statistics Agency (BPS). In this study data is the number of families in the district in terms of population data obtained in 2011 . Data on the population of each district in the city of Banda Aceh in 2011 can be seen from Table 2.2 below:
Tabel 3.1. Population of Banda Aceh in 2011 No.
Kecamatan
Jumlah Populasi (Jiwa)
1.
Meuraxa
16.861
2.
Jaya Baru
22.535
3.
Banda Raya
21.369
4.
Baiturrahman
31.073
5.
Lueng Bata
24.132
6.
Kuta Alam
43.184
7.
Kuta Raja
10.672
8.
Syiah Kuala
35.648
9.
Ulee Kareng
23.088
Total Populasi
228.562
Sources : Badan Pusat Statistik (2011)
2.3
Method of Analysis Data We applied a CHAID-based tree induction method to identify the rules that describe which
choices (i.e., actions) are made under which conditions. CHAID (Chi-square Automatic Interaction Detector) generates non-binary trees, i.e., trees where more than two branches can be attached to a single root or node, based on a relatively simple algorithm that is particularly well suited for the analysis of larger datasets. Other decision tree induction systems are C4.5 (Quinlan, 1993) and CART (Breiman et 67
al., 1984). CHAID relies on the Chi-square test to determine the best next split at each step. CHAID generates a decision tree by splitting subsets of the space into two or more nodes repeatedly, beginning with the entire data set (Kass, 1980). In the present context, this means that it finds the conditions that allow us to differentiate between activity participation or not for each activity type. To determine the best split at any node, it evaluates each predictor variable and merges any allowable pair of categories of that predictor variable if there is no statistically significant difference within the pair with respect to the target variable. The split that maximizes a significance value of a Chi-square test, after adjustment for multiple tests, acrossing predictor variables is used for splitting if the split is significant. The process is repeated for each newly created group until no more significant splits are found. In CHAID analysis used variables distinguished the dependent variable (the dependent variable) and the independent variable (the independent variable). The CHAID classification is based on the existing relationship between the two variables, therefore CHAID included in determining the method of dependencies within the segmentation. According to Gallagher (2000), CHAID will distinguish the independent variable variables into three different forms, namely : 1. Monotonic : the categories in this variable can be combined or merged by CHAID only if they are adjacent to each other , ie the variables that its categories follow the original sequence ( ordinal data ) , eg age or income . 2. Free: categories in this variable can be combined or merged or not , although both adjacent to each other ( nominal data ) , for example : employment , ethnic group , and geographic area 3. Floating: categories on this variable will be treated as monotonic except for the last category (missing value), which can be combined with any category .
Results in the formation of CHAID segment will be displayed in a tree diagram. This diagram is often used in modeling trip generation, for the activity classification. The most popular method used for relatively rapid development and results of the model built easily understood. Decision tree is used for data as well as more complex and very difficult to be analyzed . Decision tree is one of the classification method that uses a tree structure representation (tree) where each node attributes present, branches present value of the attribute and leaves the class present. The top node of the decision tree is called the root . In the decision tree there are 3 types of nodes , namely ; a) Root Node is a top node , the node does not exist can not have input and output or output has more than one . b) Internal node is a branching node , the node there is only one input and output have at least two . 68
c) Leaf node or a terminal node is an end node , the node there is only one input and output have not .
3.
METHODS
Table 3.1 . Number of Sampling Data Banda Aceh No.
District
1 2 3 4 5 6 7 8 9
Meuraxa Jaya Baru Banda Raya Baiturrahman Lueng Bata Kuta Alam Kuta Raja Syiah Kuala Ulee Kareng TOTAL
Number of Populasi 16,861 22,535 21,369 31,073 24,132 43,184 10,672 35,648 23,088 228,562
Sampel Responden 1,356.70 1,384.73 1,380.11 1,408.50 1,390.38 1,426.62 1,296.25 1,416.73 1,386.77 12,446.81
Sampel Families 339.17 346.18 345.03 352.12 347.60 356.66 324.06 354.18 346.69 3,111.70
In this study, the number of samples obtained from the population of the city of Banda Aceh as many families 3111.70 by assuming the number of respondents = 1 every 4 families, so that the sample is rounded to 3200 families. As for the results to be obtained from the questionnaire respondents obtained containing socioeconomic factors and land-use data processing then performed using correlation test for each variable with the help of software SPSS Decision Tree method is analyzed through CHAID . Analysis results will be obtained five types of models of seizure activity conducted with family members for non work purposes in the city of Banda Aceh for each - each type of household , the household without children , households with less than two children and households having more than two children . Types of households with children divided according to the level of the child's age households with children under 12 years old and who have children over 12 years .
REFERENCES Anggraini R, et.all, 2006. A model of within-households travel activity decisions capturing interactions between household heads, 8th International DDSS Conference, Eindhoven University of Technology. Anggraini R, et.all, 2007. Modeling household activity generation and allocation decisions using decision tree induction, Paper Presented in WCTR Conference 2007, Berkeley USA.
69
Anggraini R, 2009. Household activity-travel behavior (implementation of within-household interactions), PhD thesis, Eindhoven University of Technology. Arikunto. S. 2006. Prosedur penelitian suatu pendekatan praktek, Edisi Revisi VI, Penerbit Rineka Cipta, Jakarta Bagozzi, R.P, 1994. Advanced methods of marketing research. Blackwell Publishers Ltd, Oxford. Bruton M.J. 1985, Introduction to transportation planning, Hutchinson Technical Education, London. Gallagher,
C.A,
2000.
An
iterative
approach
tp
classification
analysis.
www.casact.org/pubs/dpp/dpp90/90dpp237.pdf. Tanggal akses : 18 Juli 2013. Lehmann,
T.
dan
Eherler,
D.
2001.
Responder
profiling
with
CHAID and
dependency
analysis.www.informatik.uni-freiburg.de/-ml/ecmlpkdd/WS-proceeding/w10/lehmann.pdf. Tanggal akses : 20 Juni 2013 McNally, M. G. and Craig R. Rindt (2007). The activity-based approach, Department of Civil and Environmental Engineering and Institute of Transportation Studies University of California. Morlok, E. K. 1995. Pengantar teknik dan perencanaan transportasi, Penerbit Erlangga Jakarta. Myers, J.H. 1996. Segmentation and positioning for strategic marketing decisions. American Marketing Association. Chicago Tamin. O.Z. 2008. Perencanaan dan pemodelan transportasi, Penerbit ITB, Bandung. Vovsha. P, Petersen. E, Donnely. R. 2004. A Model for Allocation of Maintenance Activities to the Household Members, Papers Presented at the 83th Annual Meeting of the TRB, Washington D.C.
70
Article #6 (Draft article to be submitted to Journal of Intelligent Transportation and Urban Planning) http://www.bowenpublishing.com/itup/
Implementing the Activity-based Model of Albatross in Indonesian City (Case Study of Banda Aceh City) 1
2
Renni Anggraini1, Theo A. Arentze2, Harry J.P. Timmermans2 Department of Civil Engineering, Faculty of Engineering, Syiah Kuala University Jl.Tgk. Syech Abdul Rauf no.7, 23111, Darussalam, Banda Aceh, Indonesia
Faculty of Architecture Building & Planning, Eindhoven University of Technology PO Box 513, 5600 MB, Eindhoven, the Netherlands Email:
[email protected]
Abstract As a new application in transportation planning, travel demand modelling has recently favoured the activitybased approach to examine individuals’ travel behaviour. Albatross, a learning based transportation oriented simulation system, is a comprehensive and fully operational activity-based model developed for Dutch government. Fundamentally, Albatross is developed to predict individuals travel demand in the Dutch context, which is available in software. Given that it is developed based on Dutch data, it may not match settings exactly when applied to another country, in particular to an Asian country, such as Indonesia. For example, in terms of culture, demography, geography and economics, both countries are different. Therefore, adjustments may be needed to make Albatross compatible with the system. This study attempts to implement the Albatross system to support transport planning in Indonesia. For the first case, we consider Banda Aceh city as a pilot project. By doing adjustment in some variables and data, it is expected that the operational system of transport demand for Banda Aceh city can be developed based on Albatross system. It is expected also as a benchmark of transport system operational for other cities in Indonesia. Key Words — activity-based model, Albatross, travel demand management
INTRODUCTION As a new application in transportation planning, travel demand modelling has recently favoured the activity-based approach to examine individuals’ travel behaviour. Conventional transport demand models have often been criticized for being more suitable for construction of new infrastructure instead of managing the existing travel demand. The focus is now on how to transform the transport system into a direction that could be sustainable in the long run, rather than on planning for infrastructure investment to meet new demand. For that reason, managing existing travel demand is necessary. By understanding the individual travel behaviour, transport policy can be applied to reduce traffic congestion. 71
For some decades now, transport researchers have put considerable effort into developing what is called activity-based approaches to transport demand analysis (for some overviews, see [6] - [9]). The basic idea is that travel demand is a derived demand based on people’s desire to take part in different activities. In particular, the interrelationships among different activities with respect to temporal and spatial constraints receive special attention. It means that such models treat the activities and the travel of households with respect to where and when the activities can be carried out, and how they may be scheduled, given characteristics of the households and potential opportunities, the transport networks and various institutional constraints. Albatross is one of the few existing operational activity-based models (e.g. [1] – [5]). It is a rule-based computational process model developed for the Dutch Ministry of Transportation, Public Works and Water Management. Albatross differs from other models, which use utility maximization as a key principle for modeling activity scheduling decisions. Instead, Albatross uses IF-THEN rules as a formalism to represent and predict activity-travel choice of individuals and households. The decision rules are extracted from activity diary data in the form of a decision tree by using a CHAID-based decision tree induction method. The rules predict actions in a probabilistic manner to reproduce non-systematic variance in choice behaviour. The objective of this study is to explain about Albatross model, what features it covered, and what data input it has. Further, we compare Albatross data with situational condition of Indonesia, and if necessary, adjustment needs to be done so that Albatross is compatible with Indonesian environment. The paper is organised as follows: the next section explains the overview of Banda Aceh City in terms of population and motorization. Then, we discuss the Albatross process model that is currently used for travel demand analysis in the Netherlands. A discussion of progress made in capturing within-household interactions is explained as well in this section. After this section, we explain how the model insight could support transport planning in Indonesia. The paper is wrapped up with drawing conclusions. ALBATROSS PROCESS MODEL Albatross is an acronym of A Learning-Based Transportation Oriented Simulation System. It is a computational process model of individuals’ activity-travel scheduling behaviour for travel demand analysis. Albatross is a micro-simulation model. This means that for each 72
household of a studied population the activity-ravel schedules for a designated day of individuals within that household are generated. That is to say, only the schedules of household heads are predicted. The activity scheduling process consists of four major components: (1) work activity generation (including timing, duration, location and transport mode choice for each work trip), (2) other fixed activity generation (including timing, duration and location), (3) flexible activity generation (including timing, duration and location), and (4) trip-chaining decisions and transport mode choice for each tour. In the existing Albatross model, interactions between persons are represented only in a limited manner. Scheduling steps are made alternately between the household heads whereby the condition of the schedule after each decision step of one person is used as condition information in the next decision step of the other person, and vice versa. Some aspects, such as activity allocation, car allocation, and joint participation in activities and travelling need joint decisions of the two household heads [2]. As the above-mentioned phasing suggests, the activity types distinguished are grouped into fixed activities and flexible activities. A fixed activity can be considered as an activity that has to be done in a particular time horizon on a regular basis, due to longer term commitments made by the individual. A flexible activity is an activity that can be done freely at any time. Examples of fixed activities are work and escorting a child to school, while most non-work activities are flexible activities.
Table 1. Activity Classification in a Household Defined by Albatross No
Activity
1 2 3
Work Business Other
4
Bring/get person
5 6
Shop-1-store Shop-n-store
7 8
Service-related Social-
Clustered Activity Work Work-related
Personal (P) or Household (HH) Level P P P HH
Task activity
HH HH HH
Non-task
P
Scope of Activities Full-time and part-time Work-related Other mandatory activity (school, etc) Drop-off/pick-up children to a certain location Daily shopping Non-daily shopping Renting movie, getting (fast) food, institutional purposes, etc Meeting friends, religions, etc 73
independent
9
10
activity
Social-joint
HH
Leisureindependent
P
Leisure-joint
HH
Touringindependent Touring-joint
P HH
Meeting relatives, social activities, religions, etc Sports, café/bar, eating out, movie, museum, library Recreational activities with children, café/bar, eating out, etc Making a tour by car, bike, or foot (eg., letting out the dog, etc)
In order to identify household-level decision making in activity scheduling and taking into account available activity data, we cluster activities into 10 activity categories as displayed in Table 1. These activities are similar to the classification used in the current Albatross model. Nevertheless, to distinguish person (P) and household (HH) level activity-participation decisions, we subdivide each non-task activity category into independent and joint activities. A task activity refers to a household task. Bring/get person, shop-1-store, shop-n-store, and service-related activities (see Table 1) are considered as task activities. A non-task activity is just as a task activity that can be conducted anytime by any person in the household either independently or jointly. Social, leisure and touring activities (see Table 1) are considered as non-task discretionary activities. As said, also task activities possibly can be done jointly. In order to better capture within-household interactions, we identify the facets of activitytravel behaviour of the two household heads that require household-level decision making and expand the household activity-travel scheduling process regarding several components. Each stage is treated both at person-level and household-level. Only in the stage of generating secondary fixed activities it is done at person-level. The procedures are as follows: 1. The first component still generates a work activity pattern consisting of at most two work episodes for each person, as in the existing process model. At the person-level, it involves the choice of number of work episodes and for each work episode the choice of a start time, duration and location. The decision of allocating the car to a work activity, in particular in car-deficient households, lies at the household-level. Having allocated the car, if necessary, the household heads face the decision of choosing the transport mode to the work place which lies at person-level. 2. The second component ascertains the part of the schedule related to secondary fixed activities, including business and other mandatory activities. Different from other components, this component only assumes decisions at the person-level, as these activities do not concern household tasks or activities that are considered jointly. It determines which types of secondary fixed activities are conducted that day, the number of episodes and for each activity the start time, duration and location. Additionally, it identifies the potential triplinkage of each activity to the work activity. 3. The next component deals with household task activities and non-task activities. At the household-level, it determines activity selection of household task activities (bring/get, shop1-store, shop-n-store, service) and joint non-task activities (social-joint, leisure-joint, and touring-joint). Furthermore, it determines the allocation of task activities. 74
4. Timing of household task activities (an allocated activity) and non-task activities takes place at the next stage. The model defines the start time and duration of activity categories both at the household-level and person-level. 5. Having defined the timing, trip-chaining choices are made. It is noteworthy that this step is not treated differently compared to the existing model. 6. Location of household task activities and non-task activities is determined in the next stage. The choices are conducted either at household or person level. 7. Car allocation for non-work tour is specified at this stage, which is input to the transport mode choice model in the next stage. 8. The transport mode for each non-work tour is specified. As in the earlier stages, these choices are also conducted either at household or person level depending on whether the tour includes a joint activity or not. It is essential to note that in all decision-tree models, the results of earlier decisions are used as conditions for each next decision and that the process result is a complete schedule for each person. The joint activity participation and allocation of household tasks has already been implemented at the household level [1]. Having generated household task activities - such as bring/get person, shopping, and service-related activities – the model subsequently determines the allocation of each scheduled activity of these categories either to the male alone, the female alone, or to both heads. As expected, the female gets more responsibilities in performing household tasks, in the sense that women face different circumstances than men in taking care of household tasks, in particular if young children are present in the household. This shows that gender plays an important role in household task allocation. In terms of car allocation decision model to the work place, the model considers that the car is being allocated to either to the male, the female, or to none of the head [2]. The last option means none of the head uses car, but rather, he/she will use other mode of transport to the work place. The result reveals that the propensity of males driving to the work place is the highest among the three options considered. Women tend to use the car when men have no work activities or men’s travel time to work place is zero. This also suggests that gender still plays an important role in work commitments. SUPPORTED DATA IN ALBATROSS The data used for developing Albatross is derived from the Dutch National Travel Survey (MON=Mobiliteit Onderzoek Netherlands). The data used was collected in 2004 covering all regions in the Netherlands. The survey is conducted on a regular basis to obtain travel and activity information of residents in the Netherlands. It is a household survey and a one-day travel diary where data is collected of all household members on the diary day as well as general information about household and individual attributes such as, gender, age, vehicle ownership and driving license ownership, home location, individual income, occupation, number of working hours per week, etc. In terms of the diary, respondents are invited to give information about all trips made on a designated day as well as out-of-home activities at trip destinations. Information for each trip includes start time, trip purpose, destination, activity type at the destination, and transport mode. Situational variables and spatial geography are also reported. All in all, this survey provides an exclusive data source to analyse activity-travel behaviour of Dutch residents. 75
COMPATIBILITY DATA FOR ALBATROSS Basically, Albatross is developed to predict individuals travel demand in the Dutch context. It is available as software. Given that it is developed based on Dutch data, it may not match settings exactly when applied to another country, in particular to an Asian country, such as Indonesia. For example, in terms of culture, demography, geography and economics, both countries are different. Therefore, adjustments may be needed to make Albatross compatible with the system. Some variables included in Albatross can be taken into account without modification, such as gender, age, presence of young children, number of cars in the household, income, urban density, location accessibility, level of transport services, etc. However, classifications used should match with available data sets and circumstances. Therefore, in this section, we intend to figure out what kind of data should be taken into account for Banda Aceh City. The following data components should be considered for Banda Aceh City: 1. Individual and Household Attributes Household types As in other Indonesian cities, household type in Banda Aceh is relatively large and there are a variety of composition pattern of household members. Indeed, that is an Indonesian typical, whereas the traditional culture is kept everlasting there. Most adult people still live with their parents or their children and sometimes there are also who live with other relatives, such as grand parents, grand children, niece/nephew and auntie/uncle. Even, in some households, especially for mid - high class households, the attendance of household helper, such as babysitters, nurses, housemaids, drivers, and gardeners are also ordinary. A nuclear family household consisting of parents and children only, which is typically in European countries, is hardly seen in Indonesia. In the Albatross system, 5 household types are considered: (1) Single - non-worker (2) Single - worker (3) Double - one worker (4) Double - two workers, and (5) Double - no workers. Single means there is only one adult (age older than 18 years old) as a household head. Double indicates that there are two adults (male and female) as household heads. Since the type of household is general, then it can be used for any applications in the world. When Indonesian data is used, the household types can still be kept as it is as well. However, the existence of other household members should be taken into account as explanatory variables, such as presence of young children (currently also an attribute in Albatross), presence of relatives, presence of household helpers, and number of household members. It could be essential for prediction power of the model to add them. Age Age and gender considered in Albatross is probably as these attributes are commonly considered in many research in various area. There are five age classifications: (1) less than 35 years; (2) between 35 and 55 years; (3) between 55 and 65 years; (4) between 65 and 75 years; (5) 75 years and over. Slightly different from those classifications, for Banda Aceh population seems need to do adjustment, as age of 75 years is no longer defined as productive people. Income In terms of income, Albatross classifies it into four levels: (1) Low income (2) Low-Mid income (3) Mid-High income, and (4) High income. Hence, adjustment needs to be done 76
depending on income distribution in the city. For instance, low income standard in Dutch could be completely different from standard in Banda Aceh city. Work status In terms of employment status, there are three types being considered in Albatross: (1) nonworker (2) part-time worker and (3) full-time worker. Those who work more than 30 hours per week are considered as full-time worker. According to Central Bureau of Statistics-Indonesia [11] - called as BPS-Badan Pusat Statistik – in order to define employment status, it is primarily based on the population type, where the population is divided into two types: (1) Persons under 15 years old (not economically active) and (2) Persons 15 years old and over (economically active). Working persons mean all persons who worked for salary or assisted others in obtaining salary or profit for the duration of at least one hour during the survey week [11]. The employment environment in Indonesia is quite different from that of the Netherlands. Working hour duration is not relevance to specify employment status in Indonesia. The employment status in Indonesia is classified into: 1. Self employed without assistance of other person (s) 2. Self employed assisted by family members or temporary workers 3. Employer with permanent workers 4. Employee 5. Unpaid worker Since the employment status of Indonesian condition and Netherland condition is quite different, Albatross system need to be adjusted in this matter. Vehicle Ownerships Unlike developed countries, car is not the only private vehicle in many household in developing country cities. Motorcycle is even the most essential private vehicle instead of car, where the number of motorcycle is the highest registered vehicles in Banda Aceh. The motorcycle growth is extremely high within 5 years from 2001-2006. In Indonesia, there are some reasons for having motorcycle: (1) Low price; (2) Interesting offer from the shop; (3) Economical fuel; (4) Manoeuvrable; (5) modest maintenance The first reason indicates that motorcycle price is considerably lower than that of car. Next reason specifies that there are many offers from the motorcycle shops to attract consumer to buy. For example, the procedure of buying motorcycle is not complicated. Some motorcycle shop even does not require down payment. Another example, they offer prizes for those who buy motorcycle within a particular period. As a result, those offers attracting more people to buy motorcycle than car. The third reason, motorcycle is more efficient than that of car, because the fuel use per km is less consuming. It is also manoeuvrable, where it can move freely on highway and effortless avoiding the traffic congestion. Last reason is quite significant, since the vehicle maintenance is sometimes burdensome, motorcycle maintenance is quite low-cost and simply than that of car. Based on reasons above, therefore, some additional variables related to motorcycle should be taken into consideration as well, such as motorcycle license possession and number of motorcycles in household. In summary, the following attributes at individual and household variables ideally should be considered for Banda Aceh: 1. Household type 77
2. Age of person 3. Gender 4. Income of person 5. Work status of person 6. Presence of young children 7. Presence of relatives 8. Presence of household helpers 9. Number of household members 10. Car license possession 11. Motorcycle license possession 12. Number of cars in households 13. Number of motorcycles in households 2. Transport mode Similar to household compositions, transport mode in Indonesia is various as well. Unlike developed countries, most developing countries do not have a proper mass transportation system. Albatross only distinguishes 4 modes: (1) Car driver, (2) Car passenger, (3) Slow modes, and (4) Public Transport modes, while slow modes consist of bike and walk. The following transport modes are commonly used in Indonesian cities: 1. Car Driver 2. Car Passenger 3. Motorcycle Rider 4. Motorcycle Passenger 5. Slow (Bike and Walk) 6. Public Transport: • Bus/Minibus/Microlet • Becak/Bajaj • Ojek • Andong/Sado • Taxi • Train (only in Jakarta, KRL) Considering the transport mode alternatives listed above, it seems complicated if we have to put all alternatives in the system. It needs to be aggregated, in particular in public transport type. In addition, Albatross system also needs to be adjusted to make it compatible. 3. Accessibility measures of locations In Albatross, accessibility of locations for activities, given the home location of the household, are also measured. There are 8 variables included: 1. Daily goods sector: number of employees within 3.1 km 2. Non-daily goods sector: number of employees within 4.4 km 3. All sectors: number of employees within 4.4 km 4. Size of population within 3.1 km 5. Daily goods sector: distance within which 160 employees work 6. Non-daily goods sector: distance within which 260 employees work 7. All sectors: distance within which 4500 employees work 78
8.
Distance within which 5200 people live.
These variables are important to measures the locations accessibility, and seems need adjustment as well to suits Banda Aceh environment. 4. Other transportation measures Other transportation system measures are also included in Albatross, such as parking space and fare and public transport availability (train and bus) near home location. CONCLUSIONS Transportation problem in Banda Aceh is still in medium level and that needs to be well planned. Managing transport demand is one alternative that needs to be done in some cities in developing countries, in order to avoid more severe crowdedness. Incorporating household decision-making into activity-based models of transport demand is receiving increasing attention. Problems such as activity allocation, joint activity participation and car allocation have in the past typically been addressed separately, i.e. not in the context of a scheduling process or ad hoc as part of a more comprehensive activity-based model of transport demand. Albatross model system simulates individual and household decisions related to every facet of activity schedules generally considered relevant for activity-travel analysis. The facets include activity type, duration, start time, car allocation decisions, joint activity participation, household task allocation decisions, trip-chaining type, location, and transport mode. The system is designed as a rule-based model in which situational, household, institutional and space-time constraints as well as choice heuristics of individuals are explicitly represented in the system. It is concluded that the approach is useful for developing computational process models for forecasting travel demands. Although something needs to be adjusted, the system still relevance to be used in Indonesia, in particular to Banda Aceh City. REFERENCES [1] R. Anggraini, T.A. Arentze, and H.J.P. Timmermans, “Modeling Joint Activity Participation and Household Task Allocation”. In Proc. of AATT Conference 2008, Athens, Greece, 2008. [2] R. Anggraini, T.A. Arentze, and H.J.P. Timmermans, “Modeling Car Allocation Decisions in Automobile Deficient Households”. In Proc. of European Transport Conference 2007, Noordwijk, the Netherlands, 2007. [3] T.A. Arentze, and H.J.P. Timmermans, ALBATROSS: A Learning-based Transportation Oriented Simulation System. EIRASS, Eindhoven University of Technology, the Netherlands, 2000. [4] T.A. Arentze, and H.J.P. Timmermans, “A Learning-based Transportation Oriented Simulation System”. Transportation Research B, Vol. 38, pp. 613-633, 2004. [5] T.A. Arentze, and H.J.P. Timmermans, ALBATROSS 2.0: A Learning-based Transportation Oriented Simulation System. EIRASS, Eindhoven University of Technology, the Netherlands, 2005. [6] D. Ettema, and H.J.P Timmermans, “Theories and Models of Activity Patterns”. In Etterma, D. and Timmermans, H.J.P (eds), Activity-Based Approaches to Travel Analysis, Oxford: Pergamon, pp. 1-36, 1997. [7] M.D. McNally, “The Activity-Based Approach”. Working paper on ITS, University of California, Irvinne, USA, 2000. 79
[8] K.W. Axhausen, “Activity-based Modeling: Research Directions and Future Possibilities”. In D. Simmonds and J.J. Bates, New Look at Multi-Modal Modeling, London, Cambridge, Oxford, 2000. [9] R. Kitamura, “An Evaluation of Activity-based Travel Analysis”. Journal of Transportation, Vol. 15, pp. 9-34, 1998. [10] S. Yagi, and A. Mohammadian, “Modeling Daily Activity-Travel Tour Patterns Incorporating Activity Scheduling Decision Rules”. In Proc. the 87th Transportation Research Board, Washington D.C., USA, 2008. [11] (March, 2008) The BPS Indonesia (Central Bureau of Statistics) website. [Online] Available: http.//www.bps.go.id [12] Population Census (1971, 1980, 1990, 2000) and Supas (1995, 2005).
80
