Kobe Braynt Shot Prediction using Machine Learning

Abstract

Kobe Bryant was one of the best players of Basketball. Data regarding his 20 years played games is available on the Kaggle. We transform the categorical features by PCA and normalize the data by minmax normalization technique. Machine learning techniques such as logistic regression, Random Forest, Linear Discriminant Analysis, Naïve bayes, Gradient Boosting, Adaboost and Neural Network are applied on pre-processed data to classify whether he made shot or not.  The prediction accuracy of LR, RF, LDA, NB, GB, ABC and ANN is 67.84%, 64.22%, 67.82%, 61%, 67.8%, 68% and 67% respectively on hold an out method.  The experimental results shows that Adaboost has highest prediction accuracy as compared to others method with 5 cross validations. Finally, we have got satisfactory results as compared to our benchmark (Kaggle).