Mutual Information Feature Selection (MIFS) Based Crop Yield Prediction on Corn and Soybean Crops Using Multilayer Stacked Ensemble Regression (MSER)

Crop yield prediction model helps the farmers in order to make better decisions about the appropriate time to cultivate the crops and what types of crops to be cultivated based on environmental factors to produce better yield. Advanced ensemble regression crop yield prediction model is to predict the crop yield based on the phenotype factors includes precipitation, solar radiation, maximum temperature, minimum temperature etc. The corn and soybean crops dataset includes 38 years of yield performance data collected across 105 different locations. Comparative analysis made between correlation and mutual information on the basis of feature selection. Most related features towards crop yield and achieved better predictions on crop yield though mutual information. Our proposed mutual information based advanced ensemble regression technique involved in the prediction process of crop yield on corn and soybean crops and achieved good prediction accuracy based on phenotype factors. The predicted yield performance of advanced ensemble regression crop prediction model outperformed several supervised machine learning and advanced ensemble learning algorithms. Various regression accuracy parameter metrics such as mean absolute error, mean square error and root mean square error are also involved in performance measures. Our prediction results also ensure that weather and crop management parameters are most influential towards crop yield prediction rather than soil parameters.