Rice Quality Detection Using Gradient Tree Boosting Based On Electronic Nose Dataset

Abstract

Rice is the staple food consumed by most Indonesians. However, the quality of the rice can decline over time so that the rice becomes obsolete and cannot be consumed. For now, the traditional method to distinguish between expired rice and non-expired rice is still performed by perceiving the rice with the human's sense of smell. However, this method is considered less effective because the human sense of smell can change due to changes in body health. Therefore, we established a method for detecting the shelf life of rice by using the electronic nose dataset (e-nose). We propose a machine learning model that utilizes the e-nose to assess the quality of expired and non-expired rice. The dataset was obtained from the e-nose sensor by recording sensor information for 25 weeks and storing 1955 summaries of sensor information for seven days. Our study used the gradient tree boosting machine learning model for classification with an accuracy of 96% and an error of 4%.