RICE VARIETIES CLASSIFICATION USING MACHINE LEARNING ALGORITHMS

Abstract

Rice is one of the most important crops in the world. About one-half of the world's population is wholly dependent upon rice as food. The rice plant height is about 1.2 meters and is an annual grass. In this paper, the data set contain 5 variety of rice that is growing all over the world. The dataset contains a total of 75000 samples, of which 15000 are from each class. The data set contains 107 features from which the best 20 features are selected using Random Forest Classifier. Performance metrics such as accuracy, precision, recall, and f1 score have been compared with and without the feature selection method. Most popular machine learning algorithms, namely logistic regression, decision tree, support vector machine classifier, random forest classifier, perceptron, K-nearest neighbors’ classifier, and Gaussian naïve Bayes classifier, have been trained on 70% training - 30% testing data and 80% training - 20% testing data. Experimental results show very promising results. In random forest classification, accuracy is 99.85%, while the decision tree classifies the rice sample with 99.68% accuracy.