Sciencing Data

A short overview of what Data Science is and some of the kinds of insights we look for at Akamai.

visual data
like spice on a hot griddle
patterns emerge

Transcript

1. Sciencing Data
   Acquisition, Distillation & Consumption
   

   View Slide

2. Philip Tellis
   Principal RUM Distiller @ Akamai
   Author of the OpenSource boomerang RUM
   library
   twitter:@bluesmoon * github:@bluesmoon
   

   View Slide

3. What is Data Science?
   

   View Slide

4. A fancy term for
   Statistics?
   

   View Slide

5. A synonym for
   Computer Science?
   

   View Slide

6. A multi-disciplinary Art-Form...
   combining aspects of Statistics, Computing,
   Visual Storytelling, Domain Expertise, Typography,
   ,
   

   View Slide

7. At Akamai we primarily care about Internet
   Performance and Security
   

   View Slide

8. My team’s speciality is
   RUM
   

   View Slide

9. Search for patterns in data
   

   View Slide

10. Find abnormalities by determining what’s normal
    

    View Slide

11. Identify the parts of your site most important to your users
    

    View Slide

12. How does
    the performance of a website
    affect user behaviour?
    

    View Slide

13. OR
    ?
    

    View Slide

14. View Slide

15. View Slide

16. Does emotional state affect our perception of latency?
    

    View Slide

17. Real Data
    ●
    ●
    ●
    ●
    Sports fans can get quite emotional!
    

    View Slide

18. The US was close to a win but Canada got a last minute leg in
    

    View Slide

19. But the web audience appeared to be prescient
    

    View Slide

20. ●
    ●
    ●
    ●
    

    View Slide

21. ●
    ●
    ●
    

    View Slide

22. ●
    ●
    →
    

    View Slide

23. But what about Patience?
    

    View Slide

24. Definitions
    Load Time
    Error Tolerance
    LD
    50
    LD
    25
    

    View Slide

25. Bounce Rate v/s Page Load Time for the entire game
    

    View Slide

26. 1st & 2nd Period, Score: 0 - 0
    

    View Slide

27. 3rd Period, US Leads 2 - 0
    

    View Slide

28. End of 3rd, Canada comes back 2 - 2
    

    View Slide

29. Canada wins 3 - 2
    

    View Slide

30. patience
    cultural significance
    Correlation?
    

    View Slide

31. LD
    50
    values for different countries
    DE
    ??
    US
    5.5s
    UK
    11.5s
    AU
    ??
    CA
    13.5s
    2.5
    

    View Slide

32. The importance
    of
    Domain
    Expertise
    ● structure of a hockey
    game
    ● understanding web
    performance
    ●
    biological sciences
    

    View Slide

33. Across domains, the Statistics
    used are often the same
    

    View Slide

34. We borrow from Economics
    The Product Space
    

    View Slide

35. ...and Linguistic Anthropology
    

    View Slide

36. Navigating the Pacific
    ocean wind
    birds cloud
    star
    songs stories
    visualizations
    

    View Slide

37. Now to the boring stuff...
    

    View Slide

38. cleaned
    ●
    ● malicious
    ●
    anomalous
    Before we can analyze...
    normalized
    ●
    population
    ● time-zone
    ● currency
    

    View Slide

39. View Slide

40. Interesting People
    Data
    

    View Slide

41. ★ Understand
    ★ Borrow
    ★
    ★
    ★ story
    In Conclusion
    

    View Slide

42. Philip Tellis
    Principal RUM Distiller @ Akamai
    Author of the OpenSource boomerang RUM
    library
    twitter:@bluesmoon * github:@bluesmoon
    

    View Slide