College Students’ Physical Fitness Test Data Analysis, Visualization and Prediction Using Data Mining Techniques

Abstract

Abstract Physical Fitness test analysis is crucial in the field of kinematics and health science, the success of which can potentially help efficiently explore students’ physical issue and promote better physical education as well. Considering that the physical fitness of many college students is declining, it has become urgent to conduct a scientific, rational, and exploratory analysis of physical fitness testing data. In this paper, we visualized and explored several data mining algorithms to learn the relationship among different test items and uncover some potential patterns. Experiments are conducted on a real dataset from department of sport, Xi’an Fanyi University. The experimental results show that Catboost outperforms existing approaches in terms of prediction accuracy and F measure. The promising results will effectively evaluate students’ fitness condition, provide insights on correlation of different test items and assist educators for decision making.