Soil Evaluation for Crops Cultivation

Abstract

Our thesis has focused on cultivation of 55(fifty five) crops in 2(two) districts of Bangladesh. The classification of crop was based on the elements of soil and some other soil states. Classification algorithms we used are Random Forest, J48 Decision Tree, KNN, Neural Network, Naïve Bayes, SVM and some other algorithms. These algorithms give different % of accuracy. The main purpose of the thesis to make a proper way to find the crop which is suitable with the elements and state of a soil sample for maximum result. The main problem we faced the shortage of data and making the limited data suitable for a minimum level of accuracy for every algorithm we used. The data book was collected from Soil Research & Development Institute (SRDI), Khamarbari, Dhaka.