COTTON YIELD PREDICTION USING MACHINE LEARNING TECHNIQUES

Abstract

Predicting cotton yield is a critical task influenced by various factors, including environmental conditions, crop genotype and management practices. In yield forecast many crop models have been used in technical and strategic agricultural decision-making. The short- and long-term risk of the climate and weather was the primary cause of the expected to variability of crop production related parameters, as well as the use of natural resources. In this Research I will predict White fly attacks on cotton plants leaves that can significantly impact yield, making early detection and prediction play crucial role for effective crop management. This research proposed a novel approach combining machine learning and weather analysis to detect white fly attacks and predict their impact on cotton yield. We utilize ResNet-50 and Efficient Net models on image data to identify white fly infestations, achieving an accuracy of 98%. Additionally, I integrate weather analysis using Naive Bayes, SVM, and Logistic Regression to predict the likelihood of white fly infestations, achieving accuracies of 100%, 98.63%, and 90.21%, respectively. Our results show that the combined approach integrates cotton varieties and cultivation area with image data and weather to predict cotton yield as well and this research will help farmers to make decisions enabling farmers to take preemptive actions to mitigate losses. This research contributes to the development of precision agriculture techniques, enhancing crop resilience and productivity.