Chapter 26 - Decision-making system for crop selection based on soil

Abstract

To yield abundant crops, an agrarian has to seed the right plant, at the right time, and in the right place. The right place is determined not only by the geographic location and climate peculiarities but the type of soil as well. Soil is the mainstay of agriculture and horticulture, forming the medium in which growth and ultimately the yield of food-producing crops occur. Machine learning (ML) has emerged as a high-performance computing method to create new opportunities to unravel, quantify, and understand data-intensive processes in agricultural operational environments. ML is an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed. The goal of ML is to understand the structure of data and fit it into models so that it can be utilized by people. In this research work, we developed a prediction system for crop selection based on soil features (physical properties, chemical properties, and biological properties) with 28 total properties. Five different hypothetical dataset versions were developed as training data for machine learning algorithms. This system follows analytics maturity curve stages, that is, descriptive, predictive, and prescriptive. The system works in two parts. First, it selects the suitable crop for specific soil health. Next, the system recommends health improvement guidelines for the selected soil sample so that the most suitable crop may become more profitable. The research deals with decision tree, naïve Bayes, and random forest algorithms and helps in decision making for crop selection by predicting with the highest accuracy.