Human daily activities recognition using decision tree

Abstract

Decision tree is a supervised classifier that is easy to understand. There are various decision tree methods. This study aimed to compare the performance of decision tree methods in human activity recognition using acceleration and jerk data. The subjects performed human activity daily living, namely walking on a flat surface, walking upstairs, walking downstairs, sitting, standing, and lying down. The features were grouped into three categories: acceleration features, jerk features, and combined features of acceleration and jerk. The evaluation was done using Random Forest, J48, Logistic Model Tree, Reduced Error Pruning Tree, Decision Stump, Random Tree, and Hoeffding Tree. The results showed that Random Forest outperformed the other classifiers with acceleration features performed better than the jerk features. However, the combined acceleration and jerk features yielded the highest accuracy. In conclusion, Random Forest is the best decision tree technique in recognizing the pattern in human activity.