NBA_allstar_prediction
Almost 24000 NBA players(1980 - 2007) data(23 player stats for everyone) are collected to predict the allstar players.
Three main files to run:
allstar_prediction.m
allstar_prediction_PCA.m
allstar_prediction_selecting_Lambda.m
Three-layer neural network is trained to predict the allstar players, the accuracy is about 88%. (possible reason is allstar players are voted by people, mainly considering player stats as well as fame and so on)
PCA could be adopted to generate compressed player stats for prediction. (One data is generated by PCA with over 86% accuracy!)
Using shuffling training and CV data to predict the Lambda(L2 regularization coefficient).