Classification of activity on the human activity recognition dataset using logistic regression
Agus Eko Minarno, Wahyu Andhyka Kusuma, Rizalwan Ardi Ramandita · AIP conference proceedings · 2022
Smartphones and smartwatches have become an inevitably necessary device in everyday life for the past few years. Smartphones proliferated in the community are equipped with various sensors such as Accelerometer and Gyroscope which collect raw data. In previous studies, these sensors were placed in various positions on the human body to perform Human Activity Recognition (HAR). HAR has been widely applied to our daily lives such as for detecting health, human behavior and location tracking of health actions. The dataset used in this study utilizes data from UCI Machine Learning with 30 subjects. This study proposes the Logistic Regression method with the addition of Hyperparameters to achieve better accuracy results.