Data Science - TU/e · Data Science Betere processen en producten dankzij (Big) data Wil van der...

Data Science Betere processen en producten dankzij (Big) data

Wil van der Aalst www.vdaalst.com @wvdaalst

www.processmining.org

Data Science Center Eindhoven

http://www.tue.nl/dsce/

©Wil van der Aalst & TU/e (use only with permission & acknowledgements)

DSC/e: Competences and Research Programs 28 groups and 420+ people involved

Context: Why are we using data science, does it have the intended effect, and will

people accept it?

Analysis: How to turn data into real value (models, answers/decisions, and

visualizations/insights)?

Enabling technologies: How to get the data and deal with computational/

infrastructural challenges (big data and hard questions)?

Probability and Statistics

Stochastic Networks

Data Mining

Process Mining

Visualization

Large-Scale Distributed Systems

Data-Intensive Algorithms

Data-Driven Operations Management

Data-Driven Innovation and Business

Human and Social Analytics

Privacy, Security, Ethics, and Governance

Internet of Things

[RP1] Process Analytics: Improving Service While Cutting Costs

[RP2] Customer Journey: Correlating Events to Learn and Influence Customer Behavior

[RP3] Smart Maintenance & Diagnostics: Safeguarding Availability

[RP4] Quantified Self: Improving Performance and Well-Being

[RP5] Data Value and Privacy: Economic and Legal Aspects of Data Science

[RP6] Smart Cities: Ensuring Safety and Convenience for Citizens

[RP7] Smart Grids: Data Intensive Infrastructures

Data Science Flagship (Philips & DSC/e)

• 4 Strategic topics

• 4 TU/e departments

• 16 PhD students

• 30 Data science specialists

1. Data Driven Value Propositions

2. Healthcare Smart Maintenance

3. Optimizing Healthcare Workflows

4. Continuous Personal Health


“Data Science University” in Den Bosch


Process Mining: On the interface

between process science and

data science

As generic as a

spreadsheet!


Spreadsheet: Killer App for early computers

• VisiCalc (killer

app for Apple II,

Oct. 1979)

• Lotus 1-2-3 (killer

app for IBM PC

1983)

• Microsoft Excel

(1985)


Spreadsheet: Static data



fact derived



31 items

sold

total

value

average

distribution



How to analyze operational processes?


Process Mining: Spreadsheet for behavior

• Input: events (“things that

have happened”)

• Mandatory per event:

− case identifier

− activity name

− timestamp/date

• Optional

− resource

− transaction type

− costs

− …

case

identifier

activity

name timestamp

resource row = event



208 cases

5987 events

74 activities



batching for activities

“opstellen eindnota” and

“archiveren”


Loesje van

der Aalst

desire line

Process Discovery



process discovery

NO

modeling

needed!


event data process

model

Conformance Checking


desire line

very safe

system

Conformance Checking



conformance checking

? discovered or

hand-made




fitness of

93.5%




final inspection is

skipped 40 times




move on model

(something should have

happened, but did not)

move on log

(something happened that

should not happen)



performance analysis

average

flowtime is

1.92 months

bottleneck

NO

modeling

needed!



performance analysis

waiting time of

15.74 days

NO

modeling

needed!



animating reality

real cases

NO

modeling

needed!



16 cases are

queueing

animating reality

Deviations

Where?

Why? time

costs

…


How to get started?

• Event Data

• Process Mining Tools

• Data Science Mindset


Starting point for process mining:

Event data patient activity timestamp doctor age cost

5781 make X-ray [email protected] Dr. Jones 45 70.00

5541 blood test [email protected] Dr. Scott 61 40.00



5781 CT scan [email protected] Dr. Fox 45 1200.00

5833 surgery [email protected] Dr. Scott 24 2300.00

5781 handle payment [email protected] Carol Hope 45 0.00

5541 radiation therapy [email protected] Dr. Jones 61 140.00

5541 radiation therapy [email protected] Dr. Jones 61 140.00

… … … … … …

case id activity name timestamp other data resource


How to get started?

• Event Data




Process Mining Software


900+ plug-ins available covering the

whole process mining spectrum

©Wil van der Aalst & TU/e (use only with permission & acknowledgements) ©Wil van der Aalst & TU/e (use only with permission & acknowledgements)


How to get started?

• Event Data



Process Mining

Data Science in Action

43.000+25.000 people joined!

Starts again on October 7th 2015! Register via https://www.coursera.org/course/procmin

Conclusion

http://www.tue.nl/dsce/

Get started today! spreadsheet

for behavior

data-oriented analysis (data mining, machine learning, business intelligence)

process model analysis (simulation, verification, optimization, gaming, etc.)

performance-oriented

questions, problems and

solutions

compliance-oriented

questions, problems and

solutions

Data Science - TU/e · Data Science Betere processen en producten dankzij (Big) data Wil van der...

Documents

Transcript of Data Science - TU/e · Data Science Betere processen en producten dankzij (Big) data Wil van der...