Machine Learning based Crop Yield Prediction on Geographical and Climatic Data

Abstract

Accurate forecasts of local and regional agricultural production are essential for agricultural market contractors and farmers to assist prize agreements as early as possible in the crop growing season. Predicting the crop yield well ahead of its harvest would help farmers and market contractors strategize befitting actions to market and store their produce. These kinds of predictions will also help farmers minimize losses due to crop failure and can also help businesses that depend on agricultural products to plan their business logistics and resources. In this paper, a method is proposed which would help predict the estimate of the crop yield for a specific land based on the analysis of geographical and climatic data using Machine Learning. Regression models such as Decision Tree Regression, K-Nearest Neighbor Regression, Gaussian Process Regression and Support Vector Regression are used along with feature selection, feature scaling, cross validation and hyperparameter tuning techniques to enhance their performance.