This study aims to investigate the performance of Machine Learning (ML) techniques used in Human Activity Recognition (HAR). Techniques considered are Naïve Bayes, Support Vector Machine, K-Nearest Neighbor, Logistic Regression, Stochastic Gradient Descent, Decision Tree, Decision Tree with entropy, Random Forest, Gradient Boosting Decision Tree, and NGBoost algorithm. Following the activity recognition chain model for preprocessing, segmentation, feature extraction, and classification of human activities, we evaluate these ML techniques against classification performance metrics such as accuracy, precision, recall, F1 score, support, and run time on multiple HAR datasets. The findings highlight the importance to tailor the selection of ML technique based on the specific HAR requirements and the characteristics of the associated HAR dataset. Overall, this research helps in understanding the merits and shortcomings of ML techniques and guides the applicability of different ML techniques to various HAR datasets.
AMBATI, LOKNATH SAI and El-Gayar, Omar
"Human Activity Recognition: A Comparison of Machine Learning Approaches,"
Journal of the Midwest Association for Information Systems (JMWAIS): Vol. 2021
, Article 4.
Available at: https://aisel.aisnet.org/jmwais/vol2021/iss1/4