Human Activity Classification Using Basic Machine Learning Models

Human activity recognition (HAR) is the object of interest for many researchers in machine learning. In principle, providing accurate and reasonable information on an individual’s activities and movements for pervasive computing is a very challenging problem. Recent advances in HAR have led to advanced tracking of highly complex human behaviors. This is progressively driving humans and computers to become seamlessly integrated through devices and software. The impact of this type of research has numerous applications in different sectors. This paper presents our initial experiments on evaluating the performance of popular machine learning algorithms in predicting human behaviors accurately. Our experiments suggest that some models can accomplish high recognition accuracy and low computational cost.