Machine Learning-Based Analysis of Paddy Crop Conditions

Abstract

Malaysia, heavily reliant on rice as a staple food, faces challenges in ensuring sufficient supply due to the persistent issue of plant diseases affecting productivity. Despite being the 22nd largest rice producer in Asia, the country imports 30 to 40 percent of its annual consumption, totaling 2.7 million tonnes. While Kedah and Perlis contribute significantly to local production, overall output falls short of meeting demand. The government aims to enhance productivity for self-sufficiency and cost reduction. Plant diseases, including brown spots and leaf blasts, hinder rice growth, leading to yield loss. Current manual detection methods prove costly, inefficient, and prone to errors. A shift toward innovative, automated solutions is imperative to address these challenges and secure the stability of Malaysia's rice supply. This research will apply three machine learning algorithms which are support vector machine (SVM), logistic regression (LR) and random forest (RF) to predict the paddy conditions based on the physical appearances. The result shows that the RF has better performance on the accuracy score of 83%