Comparative Analysis of Classification Algorithms using UCI Data Set based on Smartphone Accelerometer and Gyroscope

Abstract

Mobile phone based activity recognition uses data obtained from embedded sensors to infer user’s physical activities. Therefore, many mobile phones have been equipped with sensors to enable the delivery of advanced features to the users. Accelerometer and gyroscope are the sensors that embedded to several types of mobiles devices. In this paper, we apply 17 classifier algorithms to select the best performance ones using UCI data sets. These dataset are labeled twelve human activities. To test the performance accuracy of these algorithms, the 10-fold cross validation is done using Weka 3.6.11 data mining tool. The overall accuracy rates for classifiers are exceeded 85% and nearly 96% which are encouraged results. Thus, we select the appropriate classifier algorithms based on these accuracy results to be used for online human activity recognition.