An effective crop recommendation method using machine learning techniques

Abstract

The soil is a critical component in agriculture. Soil testing is the initial step to determine the right amount of nutrients and grow particular crops in soil. Machine learning (ML) classification techniques can recommend crops based on soil nutrients. In this paper, the Wrapper-PART-Grid approach for suitable crop suggestions is proposed by incorporating data on crop recommendations. It is a hybrid approach that combines the grid search (GS) method for hyperparameter optimization with the wrapper feature selection strategy and the PART (Partial C4.5 decision tree) classifier for crop recommendation. Proposed method is compared to other machine learning techniques, i.e., Multilayer perceptron (MLP), Instance-based learning with parameter k (IBk), C4.5 decision tree (CDT), and reduced error pruning (REP) tree. True positive rate, false positive rate, precision, recall, F1-score, root mean squared error (RMSE) and mean absolute error (MAE) values are used to assess these models. Compared to the other ML models, the results of the suggested method are reliable, accurate, and most effective for advising crops. This method achieved the highest accuracy of 99.31%. In this paper, authors proposed ML-based crop recommendation technique to help farmers improve their knowledge of growing appropriate crops, reducing overall wastage and increasing yield and crop quality.