A machine learning-based crop recommendation system using soil ecology for sustainable agriculture

Agriculture fundamentally depends on comprehending the complex interplay between crops and their environmental conditions, especially soil ecology. As in India, nearly 60% of people are dependent on agriculture in rural areas, and it also supports related industries like textiles, food processing, and ag machinery by providing raw materials. As in the past, machine learning methodologies are used for various soil ecologies to deliver accurate crop suggestions. Our methodology determines the optimal crops for certain places by evaluating essential soil properties, including pH, nitrogen (N), phosphorus (P), potassium (K) levels, and organic carbon content, in conjunction with weather conditions such as temperature, rainfall, and humidity. The suggested method assesses various machine learning algorithms, including decision trees, random forests, support vector machines, and ensembles, to choose the most effective models for precise predictions. We achieved a top accuracy of 98.4% by random forest and also 99.4% accuracy by decision tree, but with a risk of overfitting. Showing that integrating soil ecology leads to more precise and sustainable crop recommendations.