Comparison of machine learning classification algorithms based on weather variables and seed characteristics for the selection of paddy seed

Abstract

Selection of seed is very crucial for the farmers before the start of the crop season. In this study therefore, an attempt has been made to compare various machine learning (ML) classification techniques for paddy seed forecast for cultivation in three major paddy producing taluk of Madurai district, Tamil Nadu viz Thirumangalam, Peraiyur, and Usilampatti. Five machine learning classification techniques viz. K-nearest neighbour (KNN), decision tree (DT), naive bayes (NB), support vector machine (SVM), and logistic regression (LR) used in this study were compared based on weather data and seed characteristics for the better predictions of a paddy seed. Various measures were used to evaluate the algorithms, including F1-score, accuracy, precision, and recall. The findings indicated that the KNN (K-Nearest Neighbour) gave a better accuracy, precision, recall, and F1-score values of about 0.99, 0.94, 1.0, and 0.96 correspondingly.  It gave the best result of the paddy seed selection which may be helpful for the farming community in getting higher yield and profit.