Wheat Crop Yield Forecasting Using Various Regression Models

Abstract

The prediction of crop yield, particularly paddy production is a challenging task and researchers are familiar with forecasting the paddy yield using statistical methods, but they have struggled to do so with greater accuracy for a variety of factors. Therefore, machine learning methods such as Elastic Net, Ridge Regression, Lasso and Polynomial Regression are demonstrated to predict and forecast the wheat yield accurately for all India-level data. Assessment metrics such as coefficient of determination ($R^{2}$ ), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Squared Error (MSE) and Mean Absolute Percentage Error (MAPE) are used to evaluate the performance of each developed model. Finally, while evaluating the prediction accuracy using evaluation metrics, the performance of the Polynomial Regression model is shown to be high when compared to other models that are already accessible from various research in the literature.