International Journal of Agriculture Extension and Social Development

author(s)
Chandra Sekhar Sanaboina and K Vidya Sree

abstract
This study article's main goal is to provide a reliable method of agricultural production prediction by utilizing historical crop yield data, pesticide usage, and climatic data. To make precise predictions, the proposed model has a hybrid approach and uses advanced machine learning algorithms like Random Forest Regression, Extreme Gradient Boosting (XGBoost) and K-Nearest Neighbors (KNN). The paper also discusses ensemble methods such as Random Forest and XGBoost ensemble methods or XGBoost and KNN ensemble methods to achieve predictive power and model stability. The model is trained and tested on comprehensive datasets of solar radiation, meteorological records and Indian agriculture and climate data. Key methodologies used include K-Fold cross validation for hyper-parameter tuning using GridSearchCV, feature selection methods to make the model more focused, and ensemble learning to make the model robust and reduce bias. The entire system is created using Python, making use of powerful libraries such as Sickie-learn and Tensor flow. Achieved a high predictive accuracy, and XGBoost surpassed other models by 95% Feature selection consisted of meteorological variables, pesticide use and crop yield records, which improved model performance significantly. These findings show that the suggested hybrid strategy for crop production prediction is reliable.