Document Type : Original Article
Authors
1
Department of Artificial Intelligence, College of Computer Science and IT, University of Anbar, Ramadi 31001, Iraq.
2
College of Education for Humanities, University of Anbar, Ramadi 31001, Iraq.
3
Department of Computer Sciences, College of Science, University of Al Maarif, Al Anbar,31001, Iraq.
4
Artificial Intelligence Sciences Department, College of sciences, Al-Mustaqbal University 51001, Babil, Iraq.
5
Department of Heet Education, General Directorate of Education in Anbar, Ministry of Education, Heet 31007 Anbar, Iraq.
6
Department of Computer Networking Systems, College of Computer and Information Technology, University of Anbar, Anbar 31001, Iraq.
10.24271/psr.2025.486520.1798
Abstract
Human Activity Recognition (HAR) helps in understanding human behavior through sensor data, particularly from smartphones. Despite advances in smartphone-based HAR, challenges such as high resource consumption and real-time processing remain an issue, often due to reliance on offline learning methods. This paper applied of different machine learning (ML) algorithms Support Vector Machines (SVM), Decision Trees (DT), Naive Bayes (NB), K-Nearest Neighbors (KNN), and Linear Discriminant Analysis (LDA) for human activity classification utilizing two datasets: UCI-HAR and Wild-SHARD. This work utilized k-fold cross-validation to evaluate model robustness and accuracy. The results shows that DT achieved the high accuracy of 98.35% on the UCI-HAR dataset, while KNN led to an accuracy of 96.86% on the Wild-SHARD dataset. Conversely, NB and LDA exhibited lower performance, highlighting their limited effectiveness in these scenarios. The findings underscore the significance of selecting suitable classification techniques to enhance the efficiency and accuracy of HAR systems, offering insights for future research and practical applications in real-time activity monitoring.
Keywords
Subjects
Passer Journal of Basic and Applied Sciences