Crop and Fertilizer Recommendation System Based on Soil Classification

Abstract

Agriculture forms a major occupation in countries like India. More than 75% people rely on farming for their daily wages. Hence, achieving good yield in the crops grown by farmers is the major concern. Various environmental factors have a significant impact on the crop yield. One such component that contributes majorly to the crop yield is soil. Due to urbanization and enhanced industrialization, the agricultural soil is getting contaminated, losing fertility, and hindering the crop yield. Machine Learning (ML) is employed for agricultural data analysis. The proposed ML based model aims at classifying the given soil sample datasets into four different classes, namely very high fertile, high fertile, moderately fertile, and low fertile soil utilizing support vector machine (SVM) technique. It also predicts the suitable crops that can be grown based on the class which the soil sample belongs to and suggests the fertilizers that can be used to further enhance the fertility of soil. Using proposed model, farmers can make decisions on which crop to grow based on the soil classification and decide upon the nitrogen–phosphorous–potassium (NPK) fertilizers ratio that can be used. Comparison of the SVM algorithm with k-nearest neighbor (k-NN), and decision tree (DT) has shown that SVM performed with a higher accuracy.