bayesian-win-probability
bayesian-win-probability copied to clipboard
Feature scaling
Had a thought about scaling the features to see if there was any quick roi wrt model performance. Seemed to help some after a bit of brief experimentation.
Cross entropy improved to: -0.6723110481065838
AUC score improved to: 0.9243612960483594


Love your approach btw! Had lots of fun playing with this and adapting it to my favourite local sport
Yep! Definitely worth doing.