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.

Kaggle bosch presentation material for Kaggle Tokyo Meetup #2

14,171 views

Published on

kaggle boschコンペに参加し15/1373位に入りました。
Kaggle Tokyo Meetup #2 での発表資料となります。
(http://www.slideshare.net/hskksk/kaggle-bosch の縮小版です)

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

Kaggle bosch presentation material for Kaggle Tokyo Meetup #2

  1. 1. Bosch Production Line Performance 2017/2/4 hskksk 1
  2. 2. • • • Result • • 2
  3. 3. RCO bosch production line performance 3
  4. 4. RCO • • R, Python, C++ • • 4
  5. 5. xgboost(R) fwrite_libsvm xgboost ( )R http://www.slideshare.net/hskksk/libsvm 5
  6. 6. : : 2016/8/17 - 2016/11/12 : Matthews correlation coefficient 6
  7. 7. Lx_Sy_Dz Lx_Sy_F{z-1} 7
  8. 8. 8
  9. 9. 0: 1,176,868 (99.4%) 1: 6,879 ( 0.6%) extremely imbalanced data 9
  10. 10. Result 10
  11. 11. • g_votte • tkm • hskksk( ) 11
  12. 12. LB 12
  13. 13. Public Leaderboard 13
  14. 14. Private Leaderboard 14
  15. 15. Top Ten ! 15
  16. 16. 16
  17. 17. • LB (CV ) • ( ) ↑ • xgboost dart 17
  18. 18. Feature engineering 18
  19. 19. • 25 • 3154 19
  20. 20. 1. ID • Forum magic feature 2. • 3. • 20
  21. 21. xgboost importance • = 1 • = 3 21
  22. 22. • • ID 22
  23. 23. • • ID 23
  24. 24. Station 38 • • Station 38 !! • ID Station 38 NA 24
  25. 25. ID 25
  26. 26. • bitmap ( 17017 ) • bitmap • • • • 26
  27. 27. 27
  28. 28. 28
  29. 29. 29
  30. 30. 30
  31. 31. 31
  32. 32. • • • • 32
  33. 33. • hskksk Line2 tkm Line0 33
  34. 34. • • • 3 fold 1 • MCC LB Feedback • tkm g_votte • LB Feedback 34
  35. 35. Public Private • tkm submit Public Score Private • • • • 35
  36. 36. • submit • • • • mcc • • 36
  37. 37. kaggle • • • Accuracy confusion matrix • • think more, try less 37
  38. 38. Enjoy Kaggle! 38

×