Predictive analytics on crop yield using supervised learning techniques

Abstract

Agriculture is one of Nigeria’s most important economic activities but with climate change is a threat to crop production and a significant impact on the national economy as unforeseen scenarios can cause a drop in crop yield. Machine learning algorithms are now being considered as decision support tools for crop yields prediction and weather forecasting. Maize is the crop selected in this study, and a stochastic gradient model of five popular regression algorithms was evaluated. The prediction system is written in Python programming language and linked to a web-based interface for ease of use and effectiveness. Using performance metrics, the result shows that stochastic gradient descent (SGD) performed best with lower error rates and better R2_score value of 0.98505036. This crop yield prediction system (CYPS) is able to predict the yield of the crop which will help farmers and analysts in decision-making. It will also help industries that make use of the agricultural product in strategizing the logistics of their business.