CLASSIFICATION OF DISEASES AND PESTS OF MAIZE USING MULTINOMIAL LOGISTIC REGRESSION BASED ON RESAMPLING TECHNIQUE OF K-FOLD CROSS-VALIDATION

Yulia Resti (Departement of Mathematics, Faculty of Mathematics and Natural Science, Universitas of Sriwijaya, Inderalaya, Indonesia)
Desi Herlina Saraswati (Departement of Mathematics, Faculty of Mathematics and Natural Science, Universitas Sriwijaya, Inderalaya, Indonesia)
Des Alwine Zayanti (Departement of Mathematics, Faculty of Mathematics and Natural Science, Universitas Sriwijaya, Inderalaya, Indonesia)
Ning Eliyati (Departement of Mathematics, Faculty of Mathematics and Natural Science, Universitas Sriwijaya, Inderalaya, Indonesia)

Indonesian Journal of Engineering and Science (IJES)
Vol. 3 No. 3 (2022): Table of Content

Some of the obstacles in the cultivation of maize that cause low productivity of maize yields are diseases and pests. Early detection of maize diseases and pests is expected to reduce farmer losses. A system for the early detection of diseases and pests can be created by classifying them based on digital images. This study aimed to classify maize diseases and pests using multinomial logistic regression. The model and testing resampling were based on resampling technique of k-fold cross-validation. The research data was obtained from the RGB color feature extraction process for each object in each class of diseases and pests of corn. The results showed that the classification into seven classes using five folds had an accuracy rate of 99.85%, macro precision of 98.59%, and macro recall of 98.15%.