Smartphone sensors produce high-dimensional feature vectors that can be utilized to recognize different human activities. However, the contribution of each vector in the identification process is different, and several strategies have been examined over time to develop a procedure that yields favorable results. This paper presents the latest Machine Learning algorithms proposed for human activity classification, which include data acquisition, data preprocessing, data segmentation, feature selection, and dataset classification into training and testing sets. The solutions are compared and thoroughly analyzed by highlighting the respective advantages and disadvantages. The results show that the Support Vector Machine (SVM) algorithm achieved an accuracy rate of 95%.