Analysis & Estimation of Soil for Crop Prediction using Decision Tree and Random Forest Regression Methods
Manoj Tolani, Ambar Bajpai, Arun Balodi, Sunny, Lunchakorn Wuttisittikulkij, Piya Kovintavewat
Abstract
The spatial soil analysis for the appropriate crop production is important for the maximal production. The crop production can be increased by the optimal selection of the crop for particular spatial land. Both the soil and environmental characteristics and attributes play an important role for the production maximization. The machine learning based prediction model accurately predicts the appropriate crop. Therefore, in the proposed work, the decision tree and random forest based prediction model is proposed for the crop prediction. Both the environmental attributes, i.e., Temperature, Humidity, Rainfall, and soil attributes, i.e., Nitrogen, Potassium, Phosphorous, ph levels are used for the training of the model. The R-square prediction score shows that the decision tree regression is 95.5% accurate and random forest regression shows 98.5% accuracy. The results reveal the accuracy of random forest regression model is superior with respect to the other existing regression models.