On Predicting Soccer Outcomes in the Greek League Using Machine Learning

The global expansion of the sports betting industry has brought prediction outcomes sport events into foreground scientific research. In this work, soccer outcome methods are evaluated, focusing on Greek Super League. Data analysis, including data cleaning, Sequential Forward Selection (SFS), feature engineering and augmentation is conducted. most important features used to train five machine learning models: k-Nearest Neighbor (k-NN), LogitBoost (LB), Support Vector Machine (SVM), Random Forest (RF) CatBoost (CB). For comparative reasons, best model also tested English Premier League Dutch Eredivisie, exploiting statistics from six seasons 2014 2020. Convolutional neural networks (CNN) transfer by encoding tabular images, using 10-fold cross-validation, after applying grid randomized hyperparameter tuning: DenseNet201, InceptionV3, MobileNetV2 ResNet101V2. This first time investigated in depth, providing performance between several deep models, as well other leagues. Experimental results all cases demonstrate that accurate CB, reporting 67.73% accuracy, while predictable league.