Klasifikasi Physical Activity Berbasis Sensor Accelorometer, Gyroscope, dan Gravity menggunakan Algoritma Multi-class Ensemble GradientBoost

Abstract

The current generation of smartphones is increasingly sophisticated, equipped with several sensors such as accelerometer, gravity sensor, and gyroscope that can be used to recognize human activities such as going up stairs, going down stairs, running and walking. To get information, the data will be grouped using statistical methods. The performance of statistical methods has shortcomings in classifying data because of the procedures that must be met. To cover this shortcoming, the ensemble technique is used. In this paper, we propose to apply the Multi-Class Ensemble Gradientboost algorithm to improve the performance of the logistic regression method in classifying such as climbing stairs, descending stairs, running and walking. The process of taking data using a smartphone by designing an Android-based .apk system. Then, the entire dataset was separated into training data and test data with a comparison percentage of 70:30. The results obtained show that the Multi-Class Ensemble Gradientboost algorithm succeeded in increasing the logistic regression performance by 27.93%