Applying data mining techniques to predict yield of rice in humid subtropical climatic zone of India

Agricultural crop productivity depends on various factors such as precipitation, climate, soil type and hydrology. Statistical methods and techniques can be used to assess the impact of these factors on crop production. By applying these techniques on historical climate and crop production data it may provide knowledge which can be used by farmers or industry and government stakeholders for strategies which will lead to increased crop production. The present research focuses on application of data mining techniques to extract knowledge from the historical agricultural dataset to predict rice crop yield for Kharif season of Humid Subtropical climatic zone of India. The performance evaluation of different classification techniques has been carried out in free and open source data mining software WEKA (Waikato Environment for Knowledge Analysis) as part of this research on the agricultural dataset. The experimental result provided include sensitivity, specificity, accuracy, mean absolute error (MAE), root mean squared error (RMSE), relative absolute error (RAE) and root relative squared error (RRSE). The result on the current dataset shows that J48 and LADTree(Logical Analysis of Data Tree) classifiers provide the best performance of the classifiers used in this research.