Business Processen
Introduction
Event-driven Process Chain (EPC)
Business Plan -> Business Models
Process-based Business Models
Datasharing for Large & Complex Databases
Business
Technologie
Technology Acceptance
Strategie
Structuur
Informatie / Communicatie
Business Process Modeling
Operations
DB Design
Architecture
Data Mining
Relevance
“…without a proper understanding of the business processes that need to be supported, they are doomed to fail.” “It is estimated that organizations that had the best results spent more than 40 percent of the total project time on discovery and construction of their initial model.” “But despite existing tool support, there is a notable uncertainty among practitioners about how to create process models that analysts and business professionals an easily analyze and understand.” => 7 Process Modeling Guidelines 7PMG
Ontwerp
Om een EPC en/of ERD te ontwerpen:
Event-driven Process Chains (EPC) + Mendling et al. (2008) + Experience + Business Models = “Process-based Business Model”
M$ PowerPoint LibreOffice Impress (gratis) http://www.libreoffice.org/download/ Dia (gratis) http://dia-installer.de/ GoogleDocs (gratis) http://docs.google.com/ (G account nodig)
EPC
“An EPC is an ordered graph of events and functions. It provides various connectors that allow alternative and parallel execution of processes. Furthermore it is specified by the usages of logical operators, such as OR, AND, and XOR. A major strength of EPC is claimed to be its simplicity and easy-to-understand notation. This makes EPC a widely acceptable technique to denote business processes.”
EPC Main Building Blocks
Event Function (Process) Connector (AND, OR, XOR) Organisational Unit Artefact (Information Object) Chronological flow of processes Information flow
Logical Operators
EPC Building Blocks
Case
A
B
AND
OR
XOR
1
1
1
1
1
0
2
0
1
0
1
1
3
1
0
0
1
1
4
0
0
0
0
0
Note: A OR B = (A XOR B) XOR (A AND B) => OR can always be avoided
EPC Assumptions
Events are triggered by the user or some external process
Example Guideline 1
Complaint handling process model
Redundant information => strict alternation of events and functions not necessary G1: use as few elements as possible
Example Guideline 2
XOR-connector with 6 arrows => reduce the number of arrows G2: Minimize the routing paths per element
Example Guideline 3
3 starting and 2 ending points => reduce and/or split G3: Use one start and one end event
Example Guideline 4
Referral of complaint is unstructured => use one start and one end function => every connector must be closed by a connector of the same type => archiving is modeled within each path G4: Model as structured as possible This must be executed anyway
Example Guideline 5
When a complaint is handled immediately and not referred, the procedure requires that (i) the complainant must be contacted, (ii) the complaint must be archived, and (iii) there is an optional follow-up that needs to take place. Two of the three paths leading to the OR-join need to be synchronized.
Example Guideline 6
Labels of events and functions have different style and are confusing Whether essential information is lost in this way, e.g. by not mentioning the specific form that must be used, depends on the exact purpose of the model and should be decided contextually. G6: User verb-object labels
G5: Avoid OR routing elements
This OR must be removed
Example Guideline 7
G7: Decompose a model with more than 50 elements
Opmerking 1: indien we alle hoofdprocessen in een afzonderlijk EPC voorstellen, komen we meestal uit op minder dan 50 elementen. Opmerking 2: het belangrijkste kenmerk van een hoofdproces is dat het slechts twee events heeft (start en einde). => 7PMG+1
Splitsen EPCs in hoofdprocessen
Splitsen EPCs in hoofdprocessen
Seven Process Modeling Guidelines (7PMG)
Case 1 – ART Monitoring
Antiretroviral Treatment Monitoring HIV-infected children in South Africa Optimal growth curve is not normal:
Stunted growth (undernourishment, co-infections) +165 million children under 5 years of age (26%) Once established, stunting and its effects become permanent. Leads to premature death later in life because vital organs never fully develop during childhood
G8: Splitsen van EPCs in hoofdprocessen
Case 1 - Problem
The growth measurements (height & weight) are affected by:
Environment Nutricion HIV ART
Semantic problem: how do we know whether ART is effective, if we only observe heights and weights? Data problem: how do we combine epidemiologic data (used to build the statistical model) with new observations? Processing problem: how does the system process the input data from the health care workers? (They don’t have a statistical background.)
Case 1 – Research Database
Research Data: collected from 7 sites (statistical cohorts):
Harriet Shezi (Soweto), Khayelitsha (Cape Town), Red Cross (Cape Town), Tygerberg (Cape Town), Mc Cord (Durban), Hlabisa (Kwa Zulu Natal), Rahima Moosa (Johannesburg)
WHO benchmark database Patient characteristics:
Case 1 – Collection of Research Data Funding Agency
Univ. 2 (CH)
Research DB was completed with a delay of several years.
sites
Gender: 1617 (50.4%) female Weight-for-age z-score: 702 (21.9%) z < -3; 590 (18.4%) -3 ≤ z < -2; 830 (25.9%) -2 ≤ z < -1; 1079 (33.7) z ≥ -1
Researc h Unit
Age group: 1034 (32.2%) <2yrs; 904 (28.2%) 2-5yrs; 1269 (39.6%) 510yrs
WHO
Stunted growth: 59.5% Outcome: 198 (6.2%) death; 268 (11%) loss to follow-up; 162 (6.7%)
Univ. 1 (SA)
transfer-out
Case 1 – EPC Data Entry
Case 1 – EPC Data Verification There is no need for the Health Care Worker to compute z-scores!
Case 1 – EPC Growth Trajectory Prediction
Case 1 – Data Entry – Patient Exists?
Case 1 – Data Entry – Patient Exists?
Case 1 – Data Entry – Show Spreadsheet
Incomplete! We need to change the historical data.
Case 1 – Data Entry – Append/Change data
Case 1 – Data Verification
Verification procedure computes Z-scores
User enters data
Case 1 – Growth Trajectory Prediction
Case 1 - Growth Trajectory Prediction
Z-score based on weight-for-age Benchmark (WHO) 95% CI 50% CI 50% CI
Measurements
Weight falls below expectations!
95% CI
Height falls below expectations!
Case 1 – Appending Data (Existing Patient)
Case 1 – Appending Data (Existing Patient)
Raw Data
Case 1 – Existing Patient (Prediction)
Model Information
Case 1 – New Patient
User Information
Case 1 – New Patient
Case 1 – New Patient
Case 2 - Types of brain imaging
Functional Magnetic Resonance Imaging (fMRI)
radioactively tagged glucose Injected
Electroencephalography (EEG)
Magneto-encephalo-graphy
Positron emission tomography (PET)
Radio waves Magnetic fields Measures blood flow
Case 2 - MEG
Record magnetic fields generated by brain activity
electrodes placed on the skull measures electrical brain waves
Magnetoencephalography (MEG)
Superconducting Quantum Interference Devices (SQUIDs) Measures magnetic fields produced by electrical activity of brain
Case 2 - Magnetoencephalography (MEG)
X GB
Grote hoeveelheden aan gegevens
Case 2 - How strong are magnetic fields from the brain?
MEG measures the fluctuations of frequency (Hz) and amplitude (T) of the brain magnetic signal
Case 2 - MEG measurement
BUT
=> SQUIDs
Other magnetic fields are much larger
We need very sensitive MEG sensors to pick up the brain magnetic fields
Earth Urban magnetic noise
The electrical activity of the heart, eye blinks also generate a field 2 to 3 order of magnitude larger than the signal from the brain! Noise is about a factor of 10³ to 10 6 larger than the MEG signal
MEG measurements need noise cancellation with extraordinary accuracy
Design of the SQUID Magnetic shielded room Hardware and software Averaging
Noise cancellation
Compensation coil compensates for variations in the background field
SQUID Design
1980
1995-2000
Whole-head sensors arrays which use 100 to 300 sensors at different locations
1st order axial gradiometer
This SQUID will only be sensitive to inhomogeneous changes of magnetic fields between the 2 coil Background fields will be spatially uniform
Pick-up coil picks up the signal from the brain
Shielded room Reduce the effect of external magnetic disturbances
Hardware and software Use of reference
Case 2 - Purpose of EEG/MEG
A linear combination of the reference output is subtracted from the MEG primary sensor output
With MEG, you can make (as in EEG):
Use of filters
Low-pass filter, high pass filter 50-Hz filter, etc...
Examples:
Use of specific software
Averaging of brain signals
Continuous acquisition of brain signals and study some events that appear « randomly » (Epileptic abnormalities, etc.) Evoked response: averaged MEG signals that are synchronous with an external stimulus or voluntary motor event
Experiment Rock-paper-scissors Effects of learning stimuli Neuro-marketing
Case 2 - Neuroscience & Datasharing
Currently all research labs maintain their data in local data centers NO sharing whatsoever MRC has criticized this Increasing pressure from peers to publish reproducible results Leading communities:
Case 2 - Neuroscience & Datasharing All datasets are stored at 2 locations (or more)
Bioinformatics Genetics Statistical Computing
Case 2 - Neuroscience & Datasharing Centralized storage & computing in the Cloud
Bittorrent
tracker
Computing at data location
Case 2 - Soorten oplossingen
Community-based
Bedrijf
Elk research labo heeft eigen datacenter Pooling/Sharing van infrastructuur Elke dataset wordt op tenminste 2 plaatsen bewaard Alle berekeningen gebeuren op de plaats van opslag Cloud Computing (opslag en verwerking) Uitbesteding (geen locale data centra)
Andere?
B Plan → B Model
Business Plan
Business Model
Redenen? Voordelen? Nadelen? Conceptueel Bepalend voor succes Bron van innovatie
Process-based Business Model
Legt verband tussen EPC en Business Model Actoren en hun afhankelijkheden
Actoren die eigenaar zijn van B Processen Externe Actoren
B Processen zijn gerelateerd aan Informatiestructuren en -stromen
Business Model (9 Building Blocks)
Alex Osterwalder: B Model Ontology
Customer Segments Value Proposition Key Resources Key Activities Key Relationships Delivery Channels Customer Relationships Revenue Streams Cost Structure
B Model based on Business Processes
B Model based on Business Processes
Event-driven Process Chains Business
Operations
Technologie
Technology Acceptance
Strategie
Structuur
Informatie / Communicatie
Business Process Modeling
DB Design
Data Mining
Architecture
Case 2 - Request Analysis
Case 2 - Request Analysis
Case 2 - Execute Analysis
Case 2 - Execute Analysis
Case 2 - Retrieve Analysis
Case 2 - Retrieve Analysis
Case 2 - Publish Analysis
Case 2 - Publish Analysis Where is the Publisher???
XML
Case 2 - Re-use Analysis
EPC-based ERD Informatie / Communicatie
Business Strategie
Business Process Modeling
Structuur
Architecture
DB Design
Operations
Case 2 - ERD – Contract & Queue & Sales
Technologie
Technology Acceptance
Data Mining
Case 2 - ERD – All Entities in 1 Database? n
View
Medical DB
Results
Name
Party
n
1
1
Task
n
Sale
Allowance
Price Costs
Report
Quantity 1
1
Priority
n
1
Data
n
n Name Party
n Bandwidth
Task
n
1
Sale
Allowance
1
Datacenter
n
1
Price Quantity
Priority
1
n RAM
FLOPS
Copy
Data 1
n
n Machine
n
1
Computatio n
Storage
Compute R Code
Datacenter 1
Type Retrieve
Parameters
Results
Bandwidth
Re-use
Copy n
1
n RAM
Machine
Computatio n
Data
FLOPS
R Code
Storage
Parameters
Raw Results
Compute Type Retrieve Re-use
Donald S. Le Vie, Jr. - Understanding Data Flow Diagrams
DFD Bestaat uit 4 componenten
Donald S. Le Vie, Jr. - Understanding Data Flow Diagrams
de oorsprong en/of bestemming van gegevens; soms ook externe entiteiten genoemd.
Gegevensopslagplaatsen stilstaande gegevens die de vorm van verschillende fysieke representaties aan kunnen nemen
Entiteiten
Gegevensstromen
Processen het werk of de handeling die op de gegevens worden uitgevoerd zodat deze getransformeerd, opgeslagen of verdeeld worden.
Case 2 - Request Analysis (EPC → DFD)
Case 2 - Request Analysis (EPC → DFD)
Case 2 - Request Analysis (EPC → DFD)
Case 2 - Request Analysis (EPC → DFD)
Case 2 - Execute Analysis (EPC → DFD)
Case 2 - Execute Analysis (EPC → DFD)
Conclusie?
Process-based Business Models
EPCs (zoals wij ze definiëren) bevatten “partiële” DFDs op niveau 0
Beide tonen de interactie tussen het systeem en de externe actoren DFDs van een hoger niveau kunnen nooit betrekking hebben op externe “sources” en “destinations” die niet op niveau 0 voorkomen
EPC = managerial tool
DFD (op hogere niveau’s) = ontwikkelingstool
Case 2 - Process-based Business Models
Business
Operations
Technologie
Technology Acceptance
Strategie
Structuur
Informatie / Communicatie
Business Process Modeling
DB Design
Architecture
Data Mining
Case 2 - Bedrijfsoplossing (Cloud Computing)
Zelfde oplossing als in het Community System:
Behalve voor “Execute Analysis” Andere?
Technologische Architectuur is verschillend Nieuwe Actor (Cloud Provider)
Krijgt een vergoeding via het bedrijf Kosten worden doorgerekend aan Research Labs
Case 2 - Execute Analysis (Cloud Computing)
Examen ●
●
●
●
●
●
●
Wat is het verschil tussen een EPC en een DFD? Illustreer het verband tussen een ERD en een EPC a.d.v. een voorbeeld. Illustreer het verband tussen een Business Model en een EPC a.d.v. een voorbeeld. Ontwerp een Business Model voor een dierenarts volgens de richtlijnen van A. Osterwalder. Ontwerp een EPC voor de administratie van een dierenarts. Beoordeel het getoonde EPC volgens de richtlijnen van Mendling et al. (2008). Leg de nadruk op de fouten van het getoonde EPC en beschrijf hoe men het probleem kan oplossen. Evalueer volgende uitspraak: “bij de modellering van de bedrijfsprocessen moet men steeds rekening houden met de Technology Architecture van een organisatie.” Illustreer uw antwoord met twee concrete voorbeelden.
Case 2 - Business Model & Business Processes