Ensemble Machine Learning Approaches for Crop Recommendation System Using Hybrid Feature Selection Techniques

Agriculture remains a vital sector in food production to feed the world's growing population. It provides raw materials to several industries such as textile, pharmaceuticals, and biofuels. Many depend on agriculture to meet their means of livelihood in the rural areas. Sustainable farming practices are crucial for environmental conservation and climate resilience.Our research investigates the concept of machine learning algorithms in agriculture, particularly in the optimization of crop yield through the analysis of soil nutrient levels and climatic variables. This study aims to develop a crop recommendation system based on hybrid feature selection techniques using NPK (Nitrogen, Phosphorus, Potassium) content and evolving climatic conditions. Five advanced machine learning boosting algorithms: XGBoost, Gradient Boost, AdaBoost, and CatBoost, along with a hybrid model, are used to evaluate their effectiveness. The dataset is of agricultural parameters, including NPK values, temperature, pH, rainfall, and humidity, while the yield data has 11 major agricultural crops and 10 horticultural crops. This approach demonstrates the potential to significantly improve the accuracy of crop selection decisions through robust ensemble machine learning methods, validated using comparative performance metrics. Keywords- Crop recommendation, Machine learning, Ensemble learning, Feature selection, Boosting algorithms, XGBoost, AdaBoost, CatBoost, Agriculture, Horticulture.