Optimizing Crop Selection using Deep Machine Learning Approaches

In agriculture, precise crop selection is crucial for optimizing yield and sustainability. This study presents a comprehensive crop selection process utilizing a dataset from the renowned Kaggle repository, which encompasses rainfall, temperature, and soil nutrient data from various regions in India. The dataset contains 2200 instances with 8 attributes, including soil characteristics (pH, nitrogen, potassium, and phosphorus) and climate factors (temperature, rainfall, and humidity). Based on these conditions, we applied 13 Machine Learning (ML) models and 4 modified Deep Learning (DL) based artificial neural network (ANN) models to determine the most suitable crops. Our analysis revealed that the Random Forest, Bagging, and CatBoost models performed exceptionally well, achieving accuracies above 99%. Among the ANN models, ANN1 demonstrated the highest metrics, indicating optimal performance for crop recommendation. This approach provides valuable insights for farmers, enabling them to select appropriate crops based on specific soil and climate conditions, thereby enhancing agricultural productivity and sustainability.