2022 FIBA 남자농구 아시안컵 경기결과를 활용한 머신러닝 분류 모형의 예측 성능 비교

초록

The purpose of this study is to compare predictive performance using traditional statistical methods, data mining techniques, and machine learning techniques using the box scores of the 2022 FIBA Men's Basketball Asian Cup Games. The subject of this study was a total of 72 match records among the records obtained through the official records of the 2022 FIBA Men's Basketball Asian Cup, and the outcome of the match was predicted through a total of 20 variables. Five classification models were used to predict the outcome of the men's basketball Asian Cup competition: K Nearest Neighbor (KNN), Decision Tree, Support Vector Machine (SVM), Logistic Regression, and Random Forest. For the data collection and processing of this study, the statistical program Python 3.10.1 version was used together with the library, and the results obtained are as follows. First, in the prediction results for each model, the SVM model showed the optimal prediction performance than the KNN, Decision Tree, Random Forest, and Logistic Regression models, and showed 86.67% prediction accuracy and 0.868 F1 score. Second, when the Random Forest classification model was used to predict the win/loss result of the 2022 Men's Basketball Asian Cup, overfitting occurred because the number of samples in the data set was not sufficient. Based on the results of this study, it was considered that more data by increasing the number of cases in the future can increase the accuracy of the model and reduce the possibility of overfitting. And in order to obtain more accurate prediction results, it is judged that research related to deep learning as well as machine learning is necessary.

키워드