Human Activity Classification using G-XGB

Wearable body sensors have recently gained popularity because they monitor user behaviors and vital indicators. These sensors are designed to monitor various physiological factors, such as motion, temperature, and heart rate, among others. Making sense of the data gathered by these sensors requires separating the dependent and independent features. The dependent parameters are those that body-worn sensors can read in this scenario. Examples of such elements include blood oxygen levels, heart rate, temperature, mobility, and airflow. These sensors capture data about a person's activity and health. The independent parameters in wearable body sensors are related to the dependent parameters, whereas the dependent parameters are those that the sensors can measure directly. Such characteristics include age, sex, weight, and environmental influences. For example, humidity and airflow might impair a temperature sensor's accuracy. The user's age or level of physical fitness may also influence the accuracy of a heart rate monitor. Activity monitors are used in this study to capture the actions of various people. Static and dynamic properties are derived from the activity monitor recording. The performance evaluations of the eight distinct machine-learning models that were created using the retrieved features are discussed in this paper. The Grid Search method is used to enhance the performance of the XGB model.