Human Activity Recognition (HAR) has been getting a lot of attention from both researchers and businesses in recent years. The goal of the HAR study has been to better understand how people act and figure out what they are planning from the inside. As each reading comes with a time stamp. For HAR to work, locating the pertinent temporal segments in the raw sensor data is crucial. Most HAR techniques require a lot of data modeling and feature extraction, both of which require an understanding of the domain. The paper suggests a technique for employing a person’s gait pattern to figure out how they walk. In the proposed article, we have employed ten machine-learning models for classifying gait patterns. All the machine learning models have been trained and tested within the same simulation environment to achieve good classification accuracy. Extra Tree Classifiers outperform other methods by a significant margin in terms of performance.