Optimizing Crop Selection Using Bagging and Boosting Models Based on Soil Composition and Environmental Factors

Crop selection is vital for sustainable agriculture and food security, demanding careful consideration of soil properties and environmental factors. This study presents a predictive model that forecasts optimal crops based on 12 input features, including soil composition like electrical conductivity, organic carbon, nitrogen, phosphorous, potassium, magnesium, calcium, zinc, and pH, as well as environmental conditions such as temperature, humidity, and rainfall. Leveraging advanced machine learning techniques like bagging, AdaBoost, gradient boost, and extreme gradient boost which using decision tree as a base model to develop models using a dataset featuring 21 crop types categorized into cereal crops, fruit crops, vegetable crops, and commercial crops, extensive hyperparameter tuning enhances model performance. The aggregation of individual models via a stacking classifier yields an ensemble model for crop selection with a remarkable accuracy of 97.77%. This high-performing model offers valuable insights for crop cultivation, tailored to soil and environmental conditions, thereby promising to bolster agricultural productivity and global food security initiatives.