An artificial intelligence-enabled crop recommendation system

Abstract

Our country is overburdened with a population of more than 160 million. So fulfilling the need for food for the whole population can be overwhelming. This research has been conducted to ensure maximum efficiency in agriculture to overcome this problem. It is a known fact that Nitrogen (N), Phosphorus (P), and Potassium (K) are the three most essential micro-nutrients of any soil. Together it is called N.P.K. Soils also contain Sulfur (S), Zinc (Zn), and Boron (B). Together we can call it S.Zn.B., which are also important micro-nutrients. Different soils have different amounts of these essentials. Based on their values, an Automated system can suggest crops for a particular land to maximize production and profitability. We propose an AI-enabled crop recommendation system that will determine the best crops based on the soil type and its N.P.K and S, Zn, and B values through a Machine Learning Model. In our research, we use a comparative analysis among some existing Machine Learning Models to identify the most efficient model for our system. This system can effectively and accurately suggest the best suitable crops for a particular land. We used Random Forest which gave us 98%, Decision Tree which gave us 98%, Naive Bayes which gave us 89%, Ensemble Model which gave us 99% of accuracy and implemented Explainable-AI. As most farmers cannot select suitable crops for their land following their soil type, the agricultural sector is facing considerable losses. To minimize this, the efficiency of crop cultivation needs to be increased. Therefore, our system can revolutionize this sector by providing effective and suitable crops for land more accurately.