Dependency analysis of various factors and ML models related to Fertilizer Recommendation

Abstract

Adopting the same fertilizer gives minimum yield to the farmers as soil properties have changed drastically due to the change in environmental condition. In literature, different algorithmic analysis has been carried out to predict the fertilizer considering various factors, however, there is a gap in identifying every possible factor relevant to fertilizer recommendation. Hence, in our proposed work, we have utilized various soil and environmental factors like Nitrogen, Phosphorus and Potassium values, humidity, rainfall, weather condition and performed a dependency analysis of these factors to give a more accurate fertilizer prediction so as to enhance the crop yield. Algorithms such as Random Forest, Decision Tree, Support Vector Machine (SVM), Naïve Bayes (NB) and Logistic Regression (LR) have been explored to study the suitability of these algorithms in fertilizer prediction. The presented algorithms are compared based on the performance metrics such as accuracy, F1 score, Recall and precision. It is found that, among other algorithms, SVM performed better with maximum accuracy of 97% when all the factors are taken into account.