features too correlated for linear regression

tylerviducic / gamescore

building a gamescore model

0 stars 0 forks source link

Open tylerviducic opened 7 months ago

tylerviducic commented 7 months ago

try ridge regression because G/xG/g/a1/a2 are too highly correlated.

tylerviducic commented 7 months ago

also try lasso regression

tylerviducic commented 7 months ago

additionally, try a grid search algorithm

tylerviducic commented 7 months ago

tried ridge regression and didn't work .

Trying grid search next.