Defining and Executing Process
Mining Workflows with
RapidProM
Ronny Mans
Wil van der Aalst
Eric Verbeek
Process Mining: Imagined Versus Real
Process
2
≠
Starting Point: Execution Data
3
SR Number Change Date+Time Status Product Owner First Name
1-364285768 2010-03-31T15:59:42+01:00 Accepted PROD582 Frederic
1-364285768 2010-03-31T16:00:56+01:00 Accepted PROD582 Frederic
1-364285768 2010-03-31T16:45:48+01:00 Queued PROD582 Frederic
1-364285768 2010-04-06T15:44:07+01:00 Accepted PROD582 Anne Claire
1-364285768 2010-04-06T15:44:38+01:00 Queued PROD582 Anne Claire
1-467153946 2011-01-31T11:18:44+01:00 Accepted PROD453 Adam
1-467153946 2011-01-31T11:19:05+01:00 Queued PROD453 Adam
1-467153946 2011-01-31T12:59:46+01:00 Accepted PROD453 Denny
1-467153946 2011-01-31T14:37:55+01:00 Accepted PROD453 Denny
1-467153946 2011-02-03T08:28:58+01:00 Queued PROD453 Denny
1-467153946 2011-02-07T12:37:33+01:00 Accepted PROD453 Paul
1-467153946 2011-02-07T12:38:25+01:00 Accepted PROD453 Paul
1-467153946 2011-03-09T11:08:06+01:00 Accepted PROD453 Åse
1-467153946 2011-03-09T11:27:05+01:00 Accepted PROD453 Åse
Extraction of Process Knowledge
4
Tool Support: ProM
PAGE 5Download from: www.processmining.org
open-source (L-GPL)
ProM
Often Encountered Issues:
• No Support for Repeating Analyses
• No Support for Scientific Experiments
• No Integration of Data Mining / Machine Learning
algorithms
PAGE 621-8-2014
Problem
No support for the
construction and execution of a workflow
which describes all
the analysis steps and their order
Solution:
RapidProM Extension
PAGE 721-8-2014
Process Mining: Basic Idea
PAGE 821-8-2014
Case Activity
1 First visit
1 MRI
2 First Visit
1 Lab test
2 MRI
1 Second visit
2 Second visit
First
visit
MRI
Lab
test
Second
visit
First
visit
MRI
Second
visit
Case 1 Case 2
First
visit
MRI
Lab
test
Second
visit
MRI
The three main types of process mining:
discovery, conformance, enhancement
PAGE 921-8-2014
Use Cases
• Discovery of the Control-flow, Organization, and
Performance Perspectives
• Selection of the Best Control-flow Model
• Decision Point Analysis
PAGE 1021-8-2014
Use Case 1
PAGE 1121-8-2014
ProM operatorsRead Log File
Use Case 1
PAGE 1221-8-2014
Analyze using
Dotted Chart
Use Case 1
PAGE 1321-8-2014
ILP Miner
Use Case 1
PAGE 1421-8-2014
Replay a Log on Petri
Net for Performance /
Conformance Analysis
Use Case 1
PAGE 1521-8-2014
Mine for Similar Task
Social Network
Use Case 1
PAGE 1621-8-2014
Use Case 2
PAGE 1721-8-2014
Process
Top PageOne Level
Down
Mine Petri Net
using Inductive
Miner
Replay a Log on Petri Net
for Conformance Analysis
Use Case 2
PAGE 1821-8-2014
Select the ‘best’ one
Use Case 3
PAGE 1921-8-2014
Based on Which Values of
Data Attributes is a Path
Chosen?
Data Attributes:
• Age
• ASA
• Urgent
ECG:
• nrInstances
Echography:
• nrInstances
Use Case 3
PAGE 2021-8-2014
Case Data ExtractorSelect AttributesSet RoleDiscretizeGuess TypesDecision TreePerformance
Use Case 3
PAGE 2121-8-2014
To Conclude
• RapidProM: Process mining for RapidMiner
• Extension available via Marketplace
Future Work:
• Incorporate techniques from other domains in
process mining analysis
• Text mining
• Semantic web
PAGE 2221-8-2014
Dr. Ronny Mans
Twitter: @ronnymans