Machine Learning based Smart Crop Recommender and Yield Predictor

Commercialization of agriculture industry raises the demand for increasing the crop yield. Crop production is highly dependent on the type, and nutrition value of soil, pH, temperature, humidity, rainfall etc. Farmers lack in determining the precise requirements of a crop, crop yield, and profit. This may cause a huge loss to farmers. Therefore, there is a need for a crop recommender system. Various researchers proposed the crop recommender systems but, these systems lack in automatic prediction of nutrient requirements, and crop yield. Also, the existing systems do not consider the export and import policies, hence fails in precise profit estimation. In this research, The authors address the above-stated challenges, and propose a machine learning based crop recommender system that predict crop yield of 22 varieties of crops. they employ machine learning models namely decision tree, light GBM, naive bayes, support vector machine, logistic, random forest, and Xgboost. The models naive bayes, random forest, and Xgboost outperforms other ML models and reports an accuracy of 99%. Furthermore, the system analyses the profit of these crops and recommend the most profitable crops to grow on a specific piece of land.