Fuzzy Decision Support System for Recommendation of Crop Cultivation based on Soil Type

Abstract

Soil with essential nutrients is capable of supporting crop cultivation. But some nutrient level in the soil is declining because of the usage of more fertilizers. Due to this, the crop production is falling. Hence, to increase the crop yield, the proposed methodology exploits all the soil micro and macronutrients of the soil to predict the crop suitability for a region. During the data categorization, beyond rough set, the fuzzy logic is used to handle the boundary values of the numerical features to improve the accuracy of the prediction. Rough set based rule induction method is used to generate the rules and the crop suitability is predicted according to the fuzzy rules. The results are benchmarked with algorithms like CN2, LEM2, AQ, and Indiscernibility. The discretized and fuzzified datasets are considered for experimental purposes. The performance of the algorithms is evaluated based on the different evaluation parameters like precision, recall, f1 score, and accuracy. The experimental results proved that fuzzy rules evolved by the LEM2 algorithm give higher prediction accuracy when compared to other algorithms for both the discretized and fuzzified datasets.