Crop Recommendation System Using Ensembling Technique

Agriculture plays a pivotal role in the Indian economy, and considered as one of predominant ancient practices. Agriculture contributes major part towards India’s GDP. There is a need to increase crop productivity. The production of a particular farm depends upon soil characteristics, environmental characteristics, but major part goes to crop selection to get a better yield. Farmers sometimes lack the knowledge to choose the best crop for their land using conventional and non-scientific methods. Incorrect crop selection can lead to loss. This work focuses on figuring out the best crop to cultivate in order to get optimum yield based on the site-specific parameters. Our proposed model takes the data of soil characteristics, environmental characteristics of a farm and the appropriate crop recommendations are given to the farmer based on the parameter values. Crop Recommendation is done through an Ensemble model using KNN, Random Forest, Gaussian Naïve Bayes, Logistic regression, SVM as base learners. To increase overall performance, the ensemble model employed in this work includes decisions from various base learners. The Majority Voting mechanism is used for combining these base learners. When compared to other methods, the results achieved with this method are more accurate. The webapp is developed to display the recommended crop when the farmer enters his farm parameters.