Improving the Classification Result of Rice Varieties Using Gradient Boosting Methods

Abstract

An accurate identification of rice grain is crucial for classifying rice varieties. This study classifies five distinct rice types that share morphological characteristics using four different machine learning methods. A total of seventy-five thousand records, consisting of fifteen thousand for each variety of rice grains, are collected from previous research. Machine learning methods that are used in this study are the Gradient Boosting method and its variances. The experimental results show that Light Gradient Boosting Machine was the algorithm with the most significant classification success rate compared to other methods, with an accuracy of 98,14%.