Human Activity Recognition Through Wearable Sensor Data: An Analytical Study

In terms of overall accuracy, the comparison shows that RBF SVM  and Logistic Regression are the strongest performers, each achieving 86.00% accuracy with balanced precision, recall, and F1-scores across most activities. The Decision Tree model follows closely with 85.00% accuracy, excelling in stationary activities like Lying, Sitting, and Standing but struggling with Jogging and Pushups. The KNN model delivers moderate results at 82.00% accuracy, showing strengths in Football and Sitting but weaknesses in Walking Downstairs and Jogging. Finally, Polynomial SVM is the weakest overall, with only 80.00% accuracy and particularly poor performance on Pushups and Jogging, despite strong recall in Football and Standing. We conclude that, if accuracy is the primary metric, RBF SVM and Logistic Regression emerge as the best models, offering the most reliable and consistent performance across the dataset.                We can explore the proposed method using various datasets and for more activities in the future. Finally, some other machine learning algorithms can be used to prove the effectiveness and we can use feature selection method.

[1]    S. M. Islam and K. H. Talukder, “Exploratory analysis of smartphone sensor data for human activity recognition,” IEEE Access, vol. 11, pp. 1–10, 2023. [2]    A. Dahou, M. A. Al-Qaness, M. Abd Elaziz, and A. M. Helmi, “MLCNNwav: Multilevel convolutional neural network with wavelet transformations for sensor-based human activity recognition,” IEEE Internet Things J., vol. 11, no. 1, pp. 820–828, 2023. [3]    U. Saha, S. Saha, M. T. Kabir, S. A. Fattah, and M. Saquib, “Decoding human activities: Analyzing wearable accelerometer and gyroscope data for activity recognition,” IEEE Sens. Lett., vol. 8, no. 1, pp. 1–4, 2024.Y. [4]    Y. Wang et al., “A novel deep multifeature extraction framework based on attention mechanism using wearable sensor data for human activity recognition,” IEEE Sens. J., vol. 23, no. 7, pp. 7188–7198, 2023. [5]    M. K. Yi, W. K. Lee, and S. O. Hwang, “A human activity recognition method based on lightweight feature extraction combined with pruned and quantized CNN for wearable device,” IEEE Trans. Consum. Electron., vol. 69, no. 3, pp. 657–670, 2023. [6]    S. Imanzadeh, J. Tanha, and M. Jalili, “Ensemble of deep learning techniques to human activity recognition using smart phone signals,” Multimed. Tools Appl., pp. 1–30, 2024. [7]    Y.-C. Lai, S.-Y. Chiang, Y.-C. Kan, and H.-C. Lin, “Coupling analysis of multiple machine learning models for human activity recognition,” Computers, Materials & Continua, vol. 79, no. 3, pp. 3783–3803, 2024, doi: 10.32604/cmc.2024.050376. [8]    D. Garcia-Gonzalez, D. Rivero, E. Fernandez-Blanco, and M. R. Luaces, “New machine learning approaches for real-life human activity recognition using smartphone sensor-based data,” Knowledge-Based Systems, vol. 279, p. 110260, 2023, doi: 10.1016/j.knosys.2023.110260. [9]    A. Logacjov, K. Bach, A. Kongsvold, H. B. Bårdstu, and P. J. Mork, “HARTH: A human activity recognition dataset for machine learning,” Sensors, vol. 21, no. 23, p. 7853, 2021, doi: 10.3390/s21237853. [10]  A. Ferrari, D. Micucci, M. Mobilio, and P. Napoletano, “Trends in human activity recognition using smartphones,” Journal of Reliable Intelligent Environments, 2021, doi: 10.1007/s40860-021-00147-0.