Prediction of Soccer Matches using Machine Learning

Abstract

The paper is concerned with predicting the result of a League and creating Strategies from gathered data using Machine Learning and Artificial Intelligence algorithms. Here we are taking data set from the real life game stats and from the massive multiplayer game series FIFA and PES. We will start by creating a web crawler to collect data-set and compare both the real world data and virtual data (Online game data) to predict the outcome of the match using supervised and unsupervised learning. Using K means clustering to segregate between different types of players such as offensive players, defensive players and goalkeepers using game data, then normalizing the virtual features to predict team strategies. We will use different models like Gaussian Naive Bayes, Hidden Markov model, Linear SVM etc. to reduce the error rates and increasing our accuracy to predict matches. This implementation can help all the teams to devise strategies for opposing team by knowing their strategies. This can also help to predict the winner of the league outcome. This prediction model can also be used in predicting Stock Market, Coaching improvements, journalism etc.