Predicting Performance in Strategic Games based on Player Habits using AI

This study looks into how machine learning can predict how well chess players perform by examining their habits, training routines, and psychological factors. We gathered data from 102 players, both amateur and professional, using a Google Form. The questions covered features such as practice hours, tactical puzzle frequency, game analysis habits, and stress levels during games. The dataset was pre-processed using label encoding for categorical variables and standardization for numerical features. We tried out three machine learning models: Random Forest, XGBoost, and Decision Tree, measuring their performance by accuracy. Among these, Random Forest achieved the highest accuracy (61.9%), followed by XGBoost (57.14%) and Decision Tree (52.38). We found that the most important factors influencing performance were the number of games played, the current FIDE rating, and the focus rating. We also created an interactive prediction tool on Google Colab, where users can enter their own data and see their performance classification. While the results demonstrate the potential of AI-driven chess training insights, the study is limited by the modest sample size (102 players) and relies on self-reported data. Future work will focus on expanding the dataset, incorporating deep learning models, and validating results on larger platforms such as lichess or chess.com. This research provides a useful framework for creating data-driven chess training strategies, which could help players, coaches, and training centers.