Process Mining: Data-driven Process
Improvement
M Faisal Reza
Lead Business Analyst, Traveloka
Process:
A series of actions or steps taken
in order to achieve a particular end
2
(Oxford Dictionaries)
Processes are Everywhere
3
Event
Activity
Decision Point
Source: Fundamentals of Business Process Management, Dumas, Da Rosa, et al
4
Variation of Business Process
● Procure to Pay: Purchasing from supplier, to paying the supplier
● Order to Cash: Order, incoming payment, packaging, delivery, revenue.
● Market to Order: Market Identification, Lead, Order, relationship
management
● Issue to Resolution
● etc.
(SAP Terminology, Source: https://wiki.scn.sap.com/wiki/display/BPX/End+to+End+Business+Scenarios)
5
Business Process Questions: e-Commerce
e-Commerce
“Which product always have a issuance problem”
“Which vendor slow response on specific product category”
“How good delivery process for vendor x”
Visit → Search → Book → Pay →Payment Verification →
→ Request for Issuance → Vendor Approve → Delivery → Goods Accepted
6
Business Process Questions: Transportation
Transportation (online ride sharing)
“How long average customer waiting time to meet driver”
“How long average driver waiting time on train station”
Search → Book → Searching for Driver → Driver Accept Booking →
→ Driver Meet Customer → Start Journey → End Journey
7
Business Process Questions: Public Services
Public Services
1. BPJS:
a. How long queue of patient in Harapan Kita hospital?
b. Which hospital have a longest waiting time for patient to be served?
c. Average patient age which queuing for more than 4 hour?
d. What segment, usually late to pay BPJS bill
2. Jasamarga & e-Money:
a. Toll Gate queue after e-Money?
8
Business Process always needs improvement
9
Improvement as a Part of Business Process Lifecycle
Source: Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016)
1
Improvement
10
● Data Availability
○ Process data is unique: Chain model, dynamics (velocity to change)
○ Several organization have event data hidden in their system.
○ But, many organization didn’t have the data or did not store properly. So it needs manual
intervention.
■ Inventory Management Six Sigma Project
● Tools & Method
Challenge on Data-driven Process Improvement
11
Design, Modelling & Enabling Improvement
What kind of data needed to store from this process? How?
Have you ever debugging code?
12
Lean Six Sigma: Statistical Process Improvement
Lean
Methodology for streamline a process by reduce “waste”.
Six Sigma
● Methodology for process improvement by reducing error/defect.
○ Increasing accuracy & consistency
○ Decreasing process variation (not only good on mean value)
● Named after a statistical concept: process only produce 3.4 defect per
million opportunities (DPMO)
13
Lean Six Sigma: Statistical Process Improvement
LSL USL 4σ
99.9937% inside quality spec
0.0063% defect opportunity
63 defect per million opportunity
- Statistical Measure
- Statistical Control
14
Now, Process Data
is Getting Bigger
Can we create Improvement?
Better and also faster?
Source:
http://www.zdnet.com/article/gartner-ibm-teradat
a-make-big-data-announcements/
15
Key Concern of Process Data
● Data is getting bigger and Bigger
● Model vs reality problem.
● Dynamic: velocity to change
● Observability: what is really happen
● Data availability
16
Exploration of solution: Process Mining
Definition:
Process Management techniques that allow for the analysis of business
process based on event logs
17
Exploration of solution: Process Mining
Event logs:
Sometimes event data hidden in the system. Log, DB Transaction, Tracking, etc
Source: Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016)
18
Exploration of solution: Process Mining
Tools:
ProM: http://www.promtools.org
BupaR: https://www.bupar.net/
PMLAB: https://github.com/josepcarmona/PMLAB
And many others, free & paid.
19
Usual Use Case
Navigation:
Auditing:
Cartography:
Source: Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016)
20
Process Discovery - Dutch Hospital
● Dutch law all hospitals need to record the
diagnostic and treatment steps at the level
of individual patients in order to receive
payments
● AMC hospital. The model was discovered
based on event data of a group of 627
gynecological oncology patients treated in
2005 and 2006.
21
Process Discovery - Dutch Hospital
● The social network is based
on handovers of work
(organizational units level).
● Flow of work between
different departments of the
AMC hospital.
● Reveals the most handovers
take place between the
gynecology department and
the general clinical lab
22
Process Discovery - Dutch Housing
Agency
● The model was discovered based on an event log
with 5987 events. All 208 cases start with activity
“010 Registreren huuropzegging” (register request
to end lease).
● Some of the activities are relatively infrequent,
e.g., activity “020 Vastleggen datum van
overlijden” occurred only 6 times (this activity is
only executed if the tenant died)
23
Usual Use Case: Auditing
24
Process Enhancement - WMO Support Request
Report & Decision
Approval
Long waiting time on Approval WMO Support request.
Social support act to assist
disabilities in The Netherland
25
Usual Use Case: Navigation
26
Key Takeaway
1. It is important to decide what kind of data, where and when to storing data to logs, so you can
observe reality of your process better.
a. Consider format of your logs.
b. Consider ideal event to storing logs. Ideally, every state changing of transaction.
2. Need to define, process or technology?
3. Process Mining are one of approach to mining insight from process data. But, not solution of all
kinds of problems, Find ideal solution for your problems.
4. Improvement of process does not mean existing process always ideal to solve your
organization problem.
5. When existing process is not possible to answer the challenge of business / demand, never
hesitate to change: Learn from the Data, Design new process
27
Reference
1. Wil van der Aalst, Process Mining - Data Science in Action (2016)
2. Dumas, Da Rosa, et al., Fundamentals of Business Process Management
(2013)
3. https://wiki.scn.sap.com/wiki/display/BPX/End+to+End+Business+Scenar
ios

Process Mining Data-driven Process Improvement - idBigdata Meetup 17 Oct 2017

  • 1.
    Process Mining: Data-drivenProcess Improvement M Faisal Reza Lead Business Analyst, Traveloka
  • 2.
    Process: A series ofactions or steps taken in order to achieve a particular end 2 (Oxford Dictionaries)
  • 3.
    Processes are Everywhere 3 Event Activity DecisionPoint Source: Fundamentals of Business Process Management, Dumas, Da Rosa, et al
  • 4.
    4 Variation of BusinessProcess ● Procure to Pay: Purchasing from supplier, to paying the supplier ● Order to Cash: Order, incoming payment, packaging, delivery, revenue. ● Market to Order: Market Identification, Lead, Order, relationship management ● Issue to Resolution ● etc. (SAP Terminology, Source: https://wiki.scn.sap.com/wiki/display/BPX/End+to+End+Business+Scenarios)
  • 5.
    5 Business Process Questions:e-Commerce e-Commerce “Which product always have a issuance problem” “Which vendor slow response on specific product category” “How good delivery process for vendor x” Visit → Search → Book → Pay →Payment Verification → → Request for Issuance → Vendor Approve → Delivery → Goods Accepted
  • 6.
    6 Business Process Questions:Transportation Transportation (online ride sharing) “How long average customer waiting time to meet driver” “How long average driver waiting time on train station” Search → Book → Searching for Driver → Driver Accept Booking → → Driver Meet Customer → Start Journey → End Journey
  • 7.
    7 Business Process Questions:Public Services Public Services 1. BPJS: a. How long queue of patient in Harapan Kita hospital? b. Which hospital have a longest waiting time for patient to be served? c. Average patient age which queuing for more than 4 hour? d. What segment, usually late to pay BPJS bill 2. Jasamarga & e-Money: a. Toll Gate queue after e-Money?
  • 8.
    8 Business Process alwaysneeds improvement
  • 9.
    9 Improvement as aPart of Business Process Lifecycle Source: Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016) 1 Improvement
  • 10.
    10 ● Data Availability ○Process data is unique: Chain model, dynamics (velocity to change) ○ Several organization have event data hidden in their system. ○ But, many organization didn’t have the data or did not store properly. So it needs manual intervention. ■ Inventory Management Six Sigma Project ● Tools & Method Challenge on Data-driven Process Improvement
  • 11.
    11 Design, Modelling &Enabling Improvement What kind of data needed to store from this process? How? Have you ever debugging code?
  • 12.
    12 Lean Six Sigma:Statistical Process Improvement Lean Methodology for streamline a process by reduce “waste”. Six Sigma ● Methodology for process improvement by reducing error/defect. ○ Increasing accuracy & consistency ○ Decreasing process variation (not only good on mean value) ● Named after a statistical concept: process only produce 3.4 defect per million opportunities (DPMO)
  • 13.
    13 Lean Six Sigma:Statistical Process Improvement LSL USL 4σ 99.9937% inside quality spec 0.0063% defect opportunity 63 defect per million opportunity - Statistical Measure - Statistical Control
  • 14.
    14 Now, Process Data isGetting Bigger Can we create Improvement? Better and also faster? Source: http://www.zdnet.com/article/gartner-ibm-teradat a-make-big-data-announcements/
  • 15.
    15 Key Concern ofProcess Data ● Data is getting bigger and Bigger ● Model vs reality problem. ● Dynamic: velocity to change ● Observability: what is really happen ● Data availability
  • 16.
    16 Exploration of solution:Process Mining Definition: Process Management techniques that allow for the analysis of business process based on event logs
  • 17.
    17 Exploration of solution:Process Mining Event logs: Sometimes event data hidden in the system. Log, DB Transaction, Tracking, etc Source: Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016)
  • 18.
    18 Exploration of solution:Process Mining Tools: ProM: http://www.promtools.org BupaR: https://www.bupar.net/ PMLAB: https://github.com/josepcarmona/PMLAB And many others, free & paid.
  • 19.
    19 Usual Use Case Navigation: Auditing: Cartography: Source:Wil van der Aalst (auth.)-Process Mining_ Data Science in Action(2016)
  • 20.
    20 Process Discovery -Dutch Hospital ● Dutch law all hospitals need to record the diagnostic and treatment steps at the level of individual patients in order to receive payments ● AMC hospital. The model was discovered based on event data of a group of 627 gynecological oncology patients treated in 2005 and 2006.
  • 21.
    21 Process Discovery -Dutch Hospital ● The social network is based on handovers of work (organizational units level). ● Flow of work between different departments of the AMC hospital. ● Reveals the most handovers take place between the gynecology department and the general clinical lab
  • 22.
    22 Process Discovery -Dutch Housing Agency ● The model was discovered based on an event log with 5987 events. All 208 cases start with activity “010 Registreren huuropzegging” (register request to end lease). ● Some of the activities are relatively infrequent, e.g., activity “020 Vastleggen datum van overlijden” occurred only 6 times (this activity is only executed if the tenant died)
  • 23.
  • 24.
    24 Process Enhancement -WMO Support Request Report & Decision Approval Long waiting time on Approval WMO Support request. Social support act to assist disabilities in The Netherland
  • 25.
  • 26.
    26 Key Takeaway 1. Itis important to decide what kind of data, where and when to storing data to logs, so you can observe reality of your process better. a. Consider format of your logs. b. Consider ideal event to storing logs. Ideally, every state changing of transaction. 2. Need to define, process or technology? 3. Process Mining are one of approach to mining insight from process data. But, not solution of all kinds of problems, Find ideal solution for your problems. 4. Improvement of process does not mean existing process always ideal to solve your organization problem. 5. When existing process is not possible to answer the challenge of business / demand, never hesitate to change: Learn from the Data, Design new process
  • 27.
    27 Reference 1. Wil vander Aalst, Process Mining - Data Science in Action (2016) 2. Dumas, Da Rosa, et al., Fundamentals of Business Process Management (2013) 3. https://wiki.scn.sap.com/wiki/display/BPX/End+to+End+Business+Scenar ios