Team-based sports have been using data analytics for decades to improve player selection, player performance, and audience engagement. Individual sports like Tennis are still lagging behind compared to other sports being individual sports mostly and unique ways of scoring. 

This project explored professional tennis match data for men and women to be applicable to a wider audience and compare and see the differences or similarities. This project utilized the tennis match data for the period when all the current professional players are active on the tour and the scope was limited to the singles matches.  

The project goal was to identify key factors that impact the match result with the available match data by developing a machine learning model to predict the match outcome. In addition, try and translate the key factors with the help of feature engineering into tangible factors for players and their coaches to incorporate into practice routines to improve their game and match performance.  

Process: 

1. Finalized the project scope. 

1. Compiled the final dataset each for Men and Women and performed exploratory analysis.  

1. Reviewed data, and created new variables.  

1. Dealt with missing values and irrelevant data. 

1. Used various machine learning techniques for model fitting. 

1. Used single and nested cross-validation to select the final models. 

1. Fine-tuned the hyperparameters to improve the model accuracy. 

1. Identified the crucial variables that impact match outcome. 

1. Summarized and compared final results with factor importance.