SOIL N-P-K PREDICTION USING LOCATION AND CROP SPECIFIC RANDOM FOREST CLASSIFICATION TECHNIQUE IN PRECISION AGRICULTURE

Abstract

Agriculture is a basic and most important profession of any country as it balances the food requirement and also the essential raw materials of industry. In the similar sense, the adaptation of implementing smart technology in agriculture practices needs to be focused on better land productivity. The fertilizer and manure should be precisely applied during agriculture practice throughout the countryside with the use of digital technologies. This motivates us to develop a reactive web application which will accurately predict the required soil N-P-K (Nitrogen-Phosphorus-Potassium) content by utilizing one time soil testing results of available soil N-P-K contents as per the yield target. This predictive model is designed by considering standard experimental data sets from Indian Council for Agriculture Research (ICAR), India. The prediction model is designed using Random Forest (RF) algorithm which is capable of handling large dataset. The predicted N-P-K content are shown in a reactive R Shiny user interface to notify required N-P-K values for the necessary action by the farmer. The complete web based prediction model is efficiently conveying the required N-P-K content information of the particular farm location to the farmer as well as the agriculture specialists.