CROP YIELD PREDICTION USING APPLIED MACHINE LEARNING AND STATISTICAL TECHNIQUES

Abstract

Accurate crop yield prediction is essential for improving agricultural planning, resource management, and food security in the face of increasing climate variability. This study proposes an integrated framework combining statistical analysis and machine learning techniques to predict crop yield using key environmental and agricultural variables, including rainfall, temperature, pesticide usage, crop type, and geographic location. A comprehensive dataset was compiled and preprocessed to ensure consistency and reliability, followed by exploratory data analysis to identify significant patterns and relationships. A Random Forest regression model was implemented due to its ability to capture nonlinear interactions and handle complex datasets. The model performance was evaluated using standard metrics, including RMSE, MAE, and R², demonstrating strong predictive accuracy and robustness. Feature importance analysis further revealed that climatic factors, particularly rainfall and temperature, are the most influential predictors of crop yield. The findings highlight the effectiveness of combining data-driven approaches with agricultural knowledge to enhance prediction accuracy. This study contributes to the development of reliable and scalable predictive systems that can support decision-making in modern agriculture.