Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Big Data, Big Flops: The gag reel of algorithms


Published on

The gag reel of algorithms. When programmers go into statistical learning without learning statistics, dangerous things can happen.

Published in: Data & Analytics
  • Be the first to comment

Big Data, Big Flops: The gag reel of algorithms

  1. 1. 4/20/16 J.S.Ramos (@xuxoramos) 1 ¡Big Data, Big Flops! The gag reel of algorithms :D
  2. 2. Radical Idea 4/20/16 J.S.Ramos (@xuxoramos) 2 “Us programmers are the worst data scientists…”.
  3. 3. The Danger Zone 4/20/16 J.S.Ramos (@xuxoramos) 3 Computer Science Math & Stats Domain Experience Danger Zone! When programmers delve into “Statistical Learning” without learning statistics. Ask the right questions Model Reality Ops Analytics Statistical Learning Data Science Predict Reality
  4. 4. Mindset IT & SW Dev Engineering HOW? Analytics Stats WHY? 4/20/16 J.S.Ramos (@xuxoramos) 4
  5. 5. Big Flop 1 4/20/16 J.S.Ramos (@xuxoramos) 5 “Famous cell phone company creates credit products for possible criminal suspects.”
  6. 6. Big Flop 2 4/20/16 J.S.Ramos (@xuxoramos) 6 “Google image classifier tags photos of 2 african- americans as ‘Gorillas’”.
  7. 7. Big Flop 3 4/20/16 J.S.Ramos (@xuxoramos) 7 “Google Flu Trends predicts influenza outbreaks based solely on searches. Nothing happens.”
  8. 8. Big Flop 4 4/20/16 J.S.Ramos (@xuxoramos) 8
  9. 9. Big Flop 5 4/20/16 J.S.Ramos (@xuxoramos) 9
  10. 10. Root cause 4/20/16 J.S.Ramos (@xuxoramos) 10 Survey of +200 data professonals. Those coming from software development have a negative correlation with business. When this red cloud turns into a deep blue oval with positive slope, Analytics will and must be born in IT.
  11. 11. How do we turn into analysts? 4/20/16 J.S.Ramos (@xuxoramos) 11 •  Hone your stats, maths and optimization skills. Start with matrix algebra. •  Read “Think Stats” by Allen Downey. •  Stop automating customs and start improving processes. •  Get closer to your business and learn its language. •  Learn R. It’s ugly, but will get you thinking in stats. •  Don’t code machine learning without learning stats!
  12. 12. 4/20/16 J.S.Ramos (@xuxoramos) 12 “Forget R vs Python. The best language for data analysis, [regardless of scale or sophistication], is the language of business!” @jokame Conclusion 1
  13. 13. 4/20/16 J.S.Ramos (@xuxoramos) 13 “Reality does not reveal itself to those who only contemplate it, but to those who immerse in it to transform it.” Octavio Paz, The Labyrinth of Solitude Conclusion 2
  14. 14. 4/20/16 J.S.Ramos (@xuxoramos) 14 Thanks! Tw: @xuxoramos LinkedIn: xuxoramos Github: jsramos