Investigations and design of predictive analytical models and Algorithms for agricultural applications

Abstract

The main aim of this research is to enhance the accuracy of crop and weather prediction from agricultural data, and to newlinedesign and develop an effective scheme for data classification. This thesis has three contributions aimed at resolving prediction issues. In experiment 1, for weather and crop prediction, the newlineagricultural data are collected from the website of the Karnataka State Department of Agriculture-(KSDA). Pre-processing is then utilized to analyse the collected data and identify the absent values of the attributes. Absent values are filled with the appropriate new values by examining the previous and past results. An advanced methodology, Weighted-Self Organizing newlineMap- (W_SOM), which is the combination of SOM and Learning Vector Quantization-(LVQ), is employed for accurate crop and weather prediction. This new approach gives a clear idea about suitable crop cultivation in the Mysore region. The experimental final results indicates that the proposed approach improves the accuracy in crop and weather prediction compared to existing approaches. newline