Authors: Shailesh Bisht, Sunny Nahar
Abstract: Agriculture is a vital sector of the Indian economy, ensuring national food security and supplying essential raw materials to various industries. As agricultural productivity becomes increasingly important in the face of climate variability and resource constraints, accurate crop yield forecasting has emerged as a critical need. This paper presents a machine learning-driven framework that leverages environmental factors such as weather conditions, soil characteristics, and the Normalized Difference Vegetation Index (NDVI) for yield prediction. The proposed system is structured into three stages: (i) forecasting weather parameters, (ii) estimating NDVI using predicted weather data, and (iii) predicting crop yield by integrating both outputs. Experiments using historical agricultural datasets demonstrate that ensemble learning techniques, particularly XGBoost and Random Forest, deliver robust performance, with XGBoost achieving the highest prediction accuracy of up to 97%.
