Machine Learning Techniques for Sugarcane Yield Prediction Using Weather Variables

Abstract

Weather has a profound influence on crop growth, development and yield. The present study deals with the use of weather parameters for sugarcane yield forecasting. Machine learning techniques like K- Nearest Neighbors (KNN) and Random Forest model have been used for sugarcane yield forecasting. Weather parameters namely maximum temperature and minimum temperature, rainfall, relative humidity in the morning and evening, sunshine hours, evaporation along with sugarcane yield have been used as inputs variables. The performance metrics like R2, Mean Square Error (MSE), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE) have been used to select the best model for predicting the yield of the crop. Among the models, Random Forest algorithm is selected as the best fit based on the high R2 and minimum error values. The results indicate that among the weather variables, rainfall and relative humidity in the evening have significant influence on sugarcane yield.