Performance Comparisson Activity Recognition using Logistic Regression and Support Vector Machine
Agus Eko Minarno, Wahyu Andhyka Kusuma, Hardianto Wibowo
Abstract
Daily activities has become leading problems in human physical analysis. Autonomous system as application in several area of human physical analysis was increase along with several machine learning methods. Fall detection, medical rehabilitation or other smart home application in physical analysis application has increase degree of life. Accelerometer and gyroscope was popular sensor for physical analysis. Several research was used these sensor with various position in human body part. Activities was separated in three class, static activity, transition activity, and dynamic activity. Basic activities has same pattern in each activity. From public HAR dataset, wich have three static activity (standing, sitting, and laying) each pattern has same shape and patterns. Dataset were used in this paper have acquire from 30 volunters. Seven basic machine learning alghoritm Logistic Regression, Support Vector Machine, Decission Tree, Random Forest, Gradien Boosted and K-Nearest Neighbor. The purposed method Logistic Regression achieves 98% accuracy same as SVM with linear kernel, with same result hyperparameter tuning for both method has same accuracy. Likewise for result where this main problems static activity was successful detected with logistic regression and SVM.