HUMAN ACTIVITY RECOGNITION BASED ON SENSOR DATA USING GRADIENT BOOSTING MODEL

Recognition of human activity focuses on identifying various human motions and behaviors using data acquired from multiple types of sensors, a branch of computer science. A branch of research focusing on environment-supported systems has taken an interest in the problem of activity and decline. This system uses a variety of technological information to monitor body movements and try to determine what activities are being done for healthcare purposes, among other applications. In this case, besides the knowledge of the artifact, the research of the fall plays a very important role. Falls are a common cause of injury and death, so it is important that the fall is diagnosed as soon as possible. This study not only provides the discovery of fall and working knowledge used in many activities in daily life, but also allows the discovery of falls when both use and practice are taken into account. Investigations using smartphone sensor data were carried out using a publicly available standardized HAR dataset called the UCI-HAR dataset, which contains activities associated with everyday life. After proper development of the data, the features are extracted with the help of feature selection techniques, followed by the support of gradient boosting classifier (xgb), which is the final step to classify the group output. The results of the study show that gradient-supported boosting outperforms previous similar methods.