Predicting Ice Hockey Results Using Machine Learning Techniques

Sport activities such as ice hockey experiences significant growth, however it is followed by lacking sports analytical techniques that enhances the success rate of ice hockey team. This study aims to analyze and compare the accuracy of various machine learning techniques in predicting ice hockey results, ultimately applying Logistic Regression to achieve higher accuracy and enhance the predictive capabilities in the sport. This study explores the prediction of field hockey scores using a number of machine learning models (ML), which were trained on data from the National Hockey League (NHL) seasons 2015-2021. The findings are presented in the form of confusion matrices, which contain number values to summarise accurate and inaccurate predictions according to class. To train the model, all of its features were carefully looked at and the most useful ones were drawn. It is shown that match-specific data, such as the features of home and away matches, is helpful to the model. The analysis result shows that the improved accuracy in predicting the outcome of ice hockey games, whereby the logistic regression model has the highest accuracy at 77.82% and therefore recommended for use in future analysis.