Andrey Novikov. Data Science in the Greenfield: How to Initiate Analytical Projects

We invite you to join the International Summer School on Data Science in Software Engineering. The summer school will be held online on 12-16 July, 2021, organized by the Laboratory of Software Testing, Tomsk Polytechnic University.
Participation is free. The official language of the school is English.

Students, young researchers and practitioners interested in applications of modern data science methods to the development and testing of complex software systems are invited to join. Follow the link to learn the full program and register your participation: https://itr-tpu.timepad.ru/event/1629835/
____
To learn more about Exactpro, visit our website https://exactpro.com/

Follow us on
LinkedIn https://www.linkedin.com/company/exactpro-systems-llc
Twitter https://twitter.com/exactpro
Facebook https://www.facebook.com/exactpro/
Instagram https://www.instagram.com/exactpro/
Vkontakte https://vk.com/exactpro_llc

Subscribe to Exactpro YouTube channel https://www.youtube.com/c/exactprosystems

Transcript

1. Data Science in Greenfield.
   How to initiate analytical
   projects.
   Andrey Novikov, Syndata.io
   

   View Slide

2. What if companies of all industries
   could make full benefit
   of data available to them?
   Syndata.io evolutionary goal
   

   View Slide

3. How do companies benefit from analytical
   projects?
   

   View Slide

4. How do companies benefit from analytical
   projects?
   • To manage it, you need to measure it.
   

   View Slide

5. How do companies benefit from analytical
   projects?
   • To manage it, you need to measure it.
   • Who is “YOU”?
   

   View Slide

6. Team for analytical projects
   • “Subject matter expert (SME)” – understands client’s realities
   and priorities. Can interpret findings.
   • Analyst – able to generate insights, build visualizations.
   

   View Slide

7. Team for analytical projects
   • “Subject matter expert (SME)” – understands client’s realities
   and priorities. Can interpret findings.
   • Analyst – able to generate insights, build visualizations.
   A project is the communication between these two.
   

   View Slide

8. How do companies benefit from analytical
   projects?
   • To manage it, you need to measure it.
   

   View Slide

9. How do companies benefit from analytical
   projects?
   • To manage it, you need to measure it.
   • What does it mean “To measure”?
   

   View Slide

10. How do companies benefit from analytical
    projects?
    • To manage it, you need to measure it.
    • What does it mean “To measure”?
    • To measure = To reduce the level of uncertainty
    

    View Slide

11. Level 1:
    Descriptive
    analytics
    

    View Slide

12. Level 1:
    Descriptive
    analytics
    Show only
    relevant info!
    On/off
    Number
    

    View Slide

13. What do we need from a dashboard?
    • Shows all relevant (and only relevant) data
    • Data updates automatically
    • Can be easily modified to support changing requirements
    

    View Slide

14. A simple Excel dashboard
    

    View Slide

15. … with drill down
    

    View Slide

16. A more advanced L1 description with Tableau
    

    View Slide

17. Even more advanced, still Level 1.
    

    View Slide

18. Who and what do you need?
    People
    • Customer / Subject
    Matter Expert
    • Analyst
    Tools
    • Excel
    • PowerBI
    • Tableau
    • Python
    Prototype gets
    embedded to
    corporate portal, ERP,
    CRM…
    

    View Slide

19. Level 2.
    Predictive analytics:
    “what will happen”,
    “what if”.
    

    View Slide

20. Level 2.
    Predictive analytics:
    “what will happen”,
    “what if”.
    Indicators come from
    Level 1 (temperature,
    humidity, rain)
    

    View Slide

21. Level 2 task examples
    • Churn prediction
    • Market campaign prediction
    • Demand prediction
    

    View Slide

22. Level 2 task examples
    • Churn prediction
    • Market campaign prediction
    • Demand prediction
    • Task duration prediction…
    When general public talks about ML, they mean L2 tasks.
    

    View Slide

23. Level 2. Who and how?
    Approaches
    • Regressions
    • Machine learning
    People
    • Analyst
    • Subj matter
    expert
    • Data Scientist
    

    View Slide

24. Level 3.
    Prescriptive
    analytics: how to
    optimize?
    

    View Slide

25. Level 3.
    Prescriptive
    analytics: how to
    optimize?
    Critical KPI for
    optimization from Level 1
    (ATM downtime, project
    duration…)
    Model from Level 2,
    predicting that KPI
    

    View Slide

26. Level 3 tasks examples
    • How to optimize project flow
    • When to change replacement parts of ATM to
    minimize downtime
    • How much food to produce
    • How long a maintenance process will take
    • …
    

    View Slide

27. Level 3. Who and how?
    Approaches
    • Machine learning
    • Optimization
    (any type)
    People
    • Analyst
    • Subj matter
    expert
    • Data Scientist
    

    View Slide

28. Customer: “Should I buy it?”
    

    View Slide

29. Project economics: Return on Analytics
    Q: Analytics is expensive.
    A: True. Resolutions:
    1) Start small with what is really important
    2) Use simple tools/techniques for PoC
    3) Run, Stop and think, Repeat.
    

    View Slide

30. Wrap up: how to initiate a meaningful project
    • Define the business-relevant focus.
    • Level 1: Slice it through numeric and visual representations
    • Level 2: Analyze for interdependencies, build prediction
    models
    • Level 3: Find room for optimization
    

    View Slide

31. Wrap up: how to initiate a meaningful project
    + Be sure to involve Subject Matter Expert to any analytics
    project you run.
    + Be reasonable at economy and use simple tools (they work).
    

    View Slide

32. Bonus:
    Several words on Tools.
    (Do not underestimate
    simple ones!)
    

    View Slide

33. PowerBI
    Good for lots of simple
    charts.
    Build fast, auto-update,
    connect to anything.
    

    View Slide

34. Excel
    Great to handover
    prototypes to
    developers.
    If you need a prototype
    and you can do it in
    Excel, then do it in Excel.
    

    View Slide

35. Python models/visualizations
    Use it on more mature
    stages, especially levels 2/3.
    

    View Slide

36. Yed.
    Visualize with graphs if you want people
    think of connections.
    

    View Slide

37. Yed.
    Visualize with graphs if you want people
    think of connections.
    

    View Slide

38. Yed.
    Yed can build graphs from adjacency
    matrices in Excel (so you can play with
    sizes, colors, forms…)
    Photos can be embedded (with some
    XML-level tricks).
    

    View Slide

39. Yed.
    Yed can build graphs from adjacency matrices in Excel
    (so you can play with sizes, colors, forms…)
    Photos can be embedded (with some XML-level tricks).
    

    View Slide

40. View Slide