Ilham A.E. Zaeni, Dyah Lestari, Soraya Norma Mustika, Dessy Rifa Anzani, Muhammad Khusairi Osman, Khairul Azman Ahmad
The present work confirms that it is possible to conduct the classification of step length using IMU sensors effectively. Data were recorded by means of accelerometers and gyroscopes; afterward, the signals were preprocessed and filtered, and features were extracted in order to enhance the accuracy of the models of classification. This study will compare three models. The model 1 used a decision tree based on four attributes and underperformed the other models. The model 2 utilized four attributes normalized by the subject's height. It achieved a more balanced performance across all classes with fewer misclassifications than the model 1. The model 3 used a Random Forest method based on the same attributes as the model 2 but normalized. It achieved an accuracy on the classification as 81.48%. The Random Forest model performed quite strongly overall but appeared to suggest that some misclassification was occurring between classes 1 and 2, therefore needing further refinement. Nonetheless, at this stage, the Random Forest model performed better among other models in regard to the balance of accuracy and error rate, hence could turn out effective for step length classification. © 2025 IEEE.
Universitas Negeri Malang, Dept. of Electrical Engineering, Malang, Indonesia; Universitas Negeri Malang, Faculty of Vocational, Malang, Indonesia; Yayasan Idetuangkan, Malang, Indonesia; Center for Electrical Eng. Studies, Universiti Teknologi MARA, Pulau Pinang, Malaysia
Research at a Glance
Register to unlock
Topics & SDG Alignment
Register to unlock
Collaboration
Register to unlock
Author Profile (Selected)
Register to unlock
References Overview
Register to unlock
Journal & Source
Register to unlock
Metadata & Integrity
Register to unlock