Solution Found using regression analysis
Job done in R, but can convert to python if wanted. See results in images. formula = PLAYER.VALUE ~ Position + Assists + Plus.Minus + Penalty.Minutes + Hits + Blocks + Faceoff.Wins Residuals: Min 1Q Median 3Q Max -0.56259 -0.23600 -0.00519 0.25032 0.53788 coefficients (Intercept) PositionC/LW PositionC/LW/RW PositionC/RW PositionD -21.25023203 -0.03265575 0.13287144 0.08317562 -15.35633084 PositionLW PositionLW/D PositionLW/RW PositionRW PositionRW/D -2.52570337 -2.78414695 -2.57518533 -8.01539925 -8.18264170 Assists Plus.Minus Penalty.Minutes Hits Blocks 1.61902054 1.10478186 0.70062691 0.33852233 0.37945591 Faceoff.Wins 0.05989731
Papan Penjelasan Umum
-
Penganjur Peraduan - 8 tahun yang lalu
Another question. Is it possible to reverse engineer this in a way without just using coefficients? Ie. finding some sort of formula that takes in to account the averages of the player stats in each position?
- 8 tahun yang lalu
-
fzvinicius
- 8 tahun yang lalu
You mean using only the averages per position? In the model I did, positions are already been considered and you can see the influence each of them have in Player Value.
Including average of each variable per position to the model does not increase the accuracy.- 8 tahun yang lalu
-
fzvinicius
- 8 tahun yang lalu
http://postimg.org/image/rygmp479r/
- 8 tahun yang lalu
-
fzvinicius
- 8 tahun yang lalu
I uploaded an image for you to see the results:
- 8 tahun yang lalu
-
Penganjur Peraduan - 8 tahun yang lalu
Hi. Ok this seems pretty nice. To be honest I don't fully understand your message, math isn't my strong point. Could you please send me the python function so I can test with some of the values? Thank you very much!
- 8 tahun yang lalu
