More than the half of the global population consume rice as their primary energy source. Therefore, this work focused on the development of a prediction model to minimize agricultural loss in the paddy field. Initially, rice plant diseases, along with their images, were captured. Then, a big data framework was used to encounter a large dataset. In this work, at first, feature extraction process is applied on the data and after that feature selection is also applied to obtain the reduced data with important features which is used as the input to the classification model. For the rice disease datasets, features based on color, shape, position, and texture are extracted from the infected rice plant images and a rough set theory-based feature selection method is used for the feature selection job. For the classification task, ensemble classification methods have been implemented in a map reduce framework for the development of the efficient disease prediction model. The results on the collected disease data show the efficiency of the proposed model.
Keywords: big data; data mining; ensemble classification; rice disease prediction; rough set theory.