A comprehensive approach on predicting the crop yield using hybrid machine learning algorithms

Abstract

Crop yield prediction is a complex task which uses historical data to predict how much yield can be obtained in a particular year. To predict accurate crop yield, a novel deep neural network named crop yield predicting deep neural network with XGBoost regression and AdaBoost regression algorithms were used. Further, in this research work, prediction models proposed are hybrid models, namely PCA-XGBoost, PCA-AdaBoost and the LSTM based Stacked Auto Encoder – Crop Yield Predicting Deep Neural Network (LSAE-CYPDNN) model for predicting the crop yield. The error metrics like Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Root Mean Squared Error (RMSE) were evaluated for the hybrid models. The result shows that the proposed hybrid LSAE-CYPDNN model yields much less MAE, MAPE and RMSE compared to the models PCA-AdaBoost and PCA-XGBoost.