인공지능을 활용한 자유형 100m 경기분석 자료의 거리 구간, 선수 및 구간내 순위에 대한 분류 예측 비교

This study was to compare the results of classified predictions on the distance categories, players, and ranks of the distance categories in 100m free style swimming using the artificial intelligence techniques. The subjects were selected and collected from the final matches of the Korea Sport Festival in 2017, 2018 and 2019. Total cases were 240 and 5 different dependent variables, such as frequencies of strokes, breaths, distance by a stroke, time by a stroke and speed, were used in this study. Among the data, 170 cases were used as training data, and 70 cases were utilized as testing data. In addition, the cluster analysis and self-organizing feature map(SOM) were used to determine data that the SOM model was trained with 300 times. There were several findings though this study as following belows. First of all, distance categories were presented with greater value of agreement between actual data and predicted result that all players performed with similar patterns in the distance categories. Secondly, players were not distinguished within this study that the utilization of SOM and cluster analysis were not suitable. It assumed that individual characteristics of performance were influenced. Thirdly, the ranks of distance categories was also not distinguished within this study that there was a limitation of utilization of those sorts of methods.