Utilização de aprendizado de máquina para previsão de resultados de jogos de futebol

Abstract

Data Mining and Machine Learning are two growing technologies that are being used in various segments of society, including commercial and academic application, among others. This work makes combines these two technologies and an widely popular sport in Brazil, soccer (or football). The purpose of this research is to use Data Mining and Machine Learning to forecast the results of soccer games and championships. In order to perform the work, were used the principles of data mining, data acquisition, pre processing,the application with machine learning algorithm swhich were chosen, Linear Regression, Naive Bayes, MLP Networks and finally SVMs, these algorithms will be responsible for the results of regression and classification. And finally the evaluation of the results, using T-Test over Accuracy and Correlation Coeficient. The results obtained were close to a case of classifiers of 60% and a correlation coefficient of approximately 0.55 in regressors. And by running the tests it was possible to get a slight advantage from SVMs and MLPs as classifiers, and a general technical drawon the regressors.