Machine Learning Approach for Crop Yield Prediction Emphasis on K-Medoid Clustering and Preprocessing

Abstract

70% of Indian population depends on farming; agriculture contributes 18% of GDP. According to government statistics 60% of crop production depends on monsoon rainfall. Hence, it is foremost important to understand the factors affecting the crop yield and there is need for development of prediction model for crop yield prediction. In this paper, we have taken Meteorological Data of Chhattisgarh (C.G.). Gathered crop production data in different districts of C.G. in last years, also collected rainfall in last years in different districts of C.G… We have proposed a machine learning model for crop yield prediction, in which dimension reduction algorithm applied to reduce the dimension of gathered data, it will suppress those data that will affect the prediction algorithm accuracy, K-medoid clustering algorithm has been applied to improve the prediction accuracy. Finally, performance of K-means clustering and K-medoid clustering is compared. For preprocessing of input dataset we have used PCA.