The purpose of this study is to find a model that explains why players leave a match and predict the likelihood of leaving a match. The data of DotA (An multiplayer battle arena game) has been analyzed. For conducting the analyses feature selection (Multiple Regression Forward Selection) and feature extraction methods (PCA, Kmeans) have been utilized. The full report of the study is brought in the Paper.PDF file.
The R file and the data are provided. For conducting the analyses using the R file you need to use Dota-Data.csv. The Raw data is provided to show the viewers how the data looked like from the beginning. The Dota-Data.csv is the processed version (after cleaning, etc.) of the Raw data.