OPTIMAL FEATURE SELECTION AND CLASSIFICATION IN CROP PREDICTION

Agriculture is a very important factor in Indian economy. The major problem faced by farmers is that they are not selecting the right crop based on parameters such as soil nutrients, humidity, water level, moisture, and seasonal weather. As a result, they are experiencing a significant loss in productivity. Machine learning algorithms are used in modern farming practices, which examine soil types as well as other factors such as weather and climatic conditions for recommending the most suitable crops. Accurate crop prediction before cultivation helps the farmer to maximize the productivity. In this work, a model is developed to predict suitable crops based on soil nutrients and other environmental factors. A machine learning framework for crop recommendation is presented using Recursive Feature Elimination (RFE) and classification. The XGBoost (Extreme Gradient Boosting) classifier is employed in this proposed system, which provides better results than other methods such as K-Nearest Neighbors, Decision Tree, Naive Bayes and Support Vector Machine. Also, the performance of RFE method is compared with Boruta and Forward Feature Selection (FFS) method. The result shows that the model with RFE based XGBoost classifier achieves high accuracy of 94%.