연관분류 마이닝 기법을 활용한 신체활동 평가

Abstract

In this paper, we analyzed the amounts of activities in target heart rate zones, i.e. ‘out-of zone’, 1 fat-burn zone’, ‘cardio zone’, and ‘peak zone’, from activity and heart rat< time-series data. Also we generated the class association rules to infer five physical activity status such as ‘inactive’, ‘sedentary’, ‘moderately active’, Vigorously active’, and ‘extremely active’. In the experiment, we evaluated the prediction power of class association rules ancverified their effectiveness by comparing classification accuracies between the proposed methoc and two benchmark methods, SVM and C4.5 decision tree model.