Process Mining: Data Science in Action - Wil van der Aalst, TU/e, DSC/e, HSE
Download as PPT, PDF9 likes2,422 views
The document discusses process mining as a data science tool to analyze organizational behavior and improve understanding of various processes. It includes examples of process discovery, conformance checking, and practical applications in educational and healthcare settings. The document emphasizes the importance of aligning actual processes with models to identify deviations and enhance efficiency.
Process Mining: Data Science in Action - Wil van der Aalst, TU/e, DSC/e, HSE
- 1. Process Mining: Data Science in Action Process Mining as a Tool to Find Out What People and Organizations Really Do prof.dr.ir. Wil van der Aalst www.processmining.org
- 2. How many students will take my class next year? PAGE 2 What is the real curriculum? When, where and why do students deviate? When, where and why do students drop out? Will I graduate or drop out? When will I graduate? Should I take this course if I want to graduate in two years from now?
- 3. PAGE 3 It's the process, stupid! (not the data, not the system, and not a particular decision)
- 4. The data scientist … PAGE 4
- 5. Data Science Center Eindhoven (DSC/e) PAGE 5 28 research groups are involved
- 6. Data Science Center Eindhoven (DSC/e) PAGE 6 7 research programs
- 12. PAGE 12 Let's play
- 13. PAGE 13 Play-Out
- 14. Play Out: A possible scenario a b d e g Case Activity Timestamp Resource 432 register travel request (a) 18-3-2014:9.15 John 432 get support from local manager (b) 18-3-2014:9.25 Mary 432 check budget by finance (d) 19-3-2014:8.55 John 432 decide (e) 19-3-2014:9.36 Sue 432 accept request (g) 19-3-2014:9.48 Mary PAGE 14
- 15. Play Out: Another scenario PAGE 15 a d c e f b d e h
- 16. Play Out: Process model allows for many more scenarios adcefcdefbdefbdeg adcefcdefbdefbdeg abdeg acdefcdefabcdbeehfbdeg abdeg adceahdbeh abdeg PAGE 16 abdeg abcefbdeh adceg adbeh acbefbdeh acdefcdefbdeh adceh adcefcdefbdefbdeg adbeh acdefcdefabcdbeehfbdeg
- 17. PAGE 17 Play-In
- 18. Play In: Simple process allowing for 4 traces abdeh abdeh abdehadbeh adbeh PAGE 18 abdeg adbeg abdeh adbeh adbeg abdeg abdehadbeh
- 19. Play In: Process allowing for more traces adcefcdefbdefbdeg adcefcdefbdefbdeg abdeg acbefbdeg adbeh adceahdbeh acdefcdefabcdbeehfbdeg PAGE 19 abdeg abcefbdeh adcegadbeh adcefcdefbdefbdeg acbefbadcedhefcdefbdeh abdeg abdeg adceh acdefcdefbdeh
- 20. No modeling needed! PAGE 20 No modeling needed!
- 21. Example Process Discovery (Dutch housing agency, 208 cases, 5987 events) PAGE 21
- 22. Example process discovery for hospital (627 gynecological oncology patients, 24331 events) PAGE 22
- 23. Process discovery algorithms (small selection) PAGE 23 heuristic mining α algorithm α# algorithm α++ algorithm distributed genetic mining language-based regions genetic mining state-based regions neural networks hidden Markov models automata-based learning stochastic task graphs conformal process graph mining block structures multi-phase mining partial-order based mining fuzzy mining LTL mining ILP mining ETM genetic algorithm Inductive Miner (infrequent)
- 24. Language based regions PAGE 24 Region R = (X,Y,c) corresponding to place pR: X = {a1,a2,c1} = transitions producing a token for pR, Y = {b1,b2,c1} = transitions consuming a token from pR, and c is the initial marking of pR.
- 25. Basic idea: enough tokens should be present when consuming A place is feasible if it can be added without disabling any of the traces in the event log. PAGE 25
- 26. PAGE 26 Replay
- 27. Replay PAGE 27 a c d e g
- 28. Replay PAGE 28 a c check budget (d) is missing! e g ?
- 29. Alignments: Relating reality and model a c » e g a c d e g check budget (d) did not happen but should have according to the model PAGE 29
- 30. Replay PAGE 30 a c h d e g reject request (h) is impossible ?
- 31. Alignments: Relating reality and model PAGE 31 a c h d e g a c » d e g reject request (h) happened but could not happen according to the model
- 32. Any trace in reality can be related to a path in the model a c e f d d b c e h PAGE 32
- 33. Any trace in reality can be related to a path in the model a c » e f d d b c e h a c d e f d » b » e h optimization problem using a cost function PAGE 33 check is missing one check too many cannot both be done
- 34. PAGE 34 process model event log synchronous move move on model only move on log only
- 35. Conformance Checking (WOZ objections Dutch municipality, 745 objections, 9583 event, f= 0.988) PAGE 35
- 36. Replay with timestamps a9.15 c9.20 d9.35 e10.15 g11.30 PAGE 36 9.15 9.20 9.35 10.15 11.30 5 55 20 40 75
- 37. Replay with timestamps for many traces PAGE 37 frequencies of activities frequencies of paths durations of activities waiting times and other delays between activities
- 38. PAGE 38 Alignments are essential! • conformance checking to diagnose deviations • squeezing reality into the model to do model-based analysis
- 39. Example: BPI Challenge 2012 (Dutch financial institute, doi:10.4121/uuid:3926db30-f712-4394-aebc-75976070e91f) PAGE 39 Work of Arya Adriansyah (Replay project)
- 40. PAGE 40 Auditor's toolbox
- 41. PAGE 41 Business analyst's toolbox
- 42. Software
- 43. 600+ plug-ins available covering the whole process mining spectrum
- 45. Disco PAGE 45
- 46. Overview: Role of process models PAGE 46 Play-Out Play-In Replay
- 48. data Big data
- 50. What if? PAGE 50 there are more than 1000 different activities? there are more than 1.000.000 cases? there are more than 100.000.000 events?
- 51. Decompose event log! vertical or horizontal PAGE 51 sets of cases sets of activities
- 52. Vertical distribution: Split cases PAGE 52 sets of cases
- 53. Horizontal distribution PAGE 53 sets of activities {a,b,e,f,g} {a,b,e,f,g} {b,c,d,e} {b,c,d,e}
- 54. Horizontal distribution: The key idea PAGE 54 projected on {a,b,e,f,g} projected on {b,c,d,e}
- 55. Two foundational ways of spitting event data: horizontal or vertical PAGE 55
- 57. Decomposing Conformance Checking PAGE 57 See "divide and conquer" framework by Eric Verbeek.
- 58. Example of a valid decomposition Log can be split in the same way! PAGE 58
- 59. Example of an alignment for observed trace a,b,c,d,e,c,d,g,f PAGE 59 Etc. a,b,c,d,e,c,d,g,f
- 60. Conformance checking can be decomposed !!! • General result for any valid decomposition: Any event log or trace is perfectly fitting the overall model if and only if it is also fitting all the individual fragments PAGE 60 for any Petri net (not just WF-nets) for any valid decompsition Wil van der Aalst, Decomposing Petri nets for process mining: A generic approach. Distributed and Parallel Databases, Volume 31, Issue 4, pp 471-507, 2013 also for data Petri nets
- 61. Example (work with Jorge Munoz-Gama and Josep Carmona) prFm6 PAGE 61
- 62. Decomposing Process Discovery PAGE 62 See "divide and conquer" framework by Eric Verbeek.
- 63. conclusion
- 64. Don't just check the temperature … PAGE 64
- 65. X-ray your processes! PAGE 65
- 66. PAGE 66 www.processmining.org www.win.tue.nl/ieeetfpm/
- 67. Process Mining: Data Science in Action https://www.coursera.org/course/procmin First Massive Open Online Course (MOOC) on Process Mining