Author Login

Correspondence to Author: S.Thavareesan^*1, J.Sriranganesan² And T. Nishatharan¹. 

¹Department of Computing, Faculty of Science, Eastern University, Sri Lanka.
² Department of Mathematics, Faculty of Science, Eastern University, Sri Lanka.

DOI: 10.52338/aast.2025.4356

Abstract:

The agricultural sector plays a critical role in South Asia’s economy, providing livelihoods for millions and ensuring food security. However, challenges such as unpredictable weather patterns, limited arable land, and increasing population pressure significantly affect crop yields. Accurate crop yield prediction is essential for addressing these issues, as it enables informed decision- making on resource allocation, crop planning, and risk management. This study evaluates the performance of various machine learning regression models for predicting crop yields in five South Asian countries: Sri Lanka, Bangladesh, India, Nepal, and Pakistan. The dataset used includes crop yield data for the ten most widely consumed crops, alongside weather-related factors like rainfall, temperature, and pesticide usage, spanning from 1961 to 2016. The models assessed include XGBoost Regressor, Decision Tree Regressor, Gradient Boosting Regressor, Random Forest Regressor, K-Nearest Neighbors (KNN), and linear models such as Linear Regression, Ridge, Lasso, Elastic Net, and Support Vector Regression (SVR). Performance was measured using Mean Squared Error (MSE) and R² scores. The results demonstrate that XGBoost Regressor achieved the lowest MSE and highest R² score, making it the most accurate model for crop yield prediction. Decision Tree Regressor and Gradient Boosting Regressor also performed well, while SVR and simpler linear models (Linear Regression and Ridge Regression) showed poorer results. These findings emphasize the effectiveness of advanced machine learning techniques, especially XGBoost, in enhancing crop yield predictions and supporting more efficient agricultural decision- making in South Asia

Keywords: Crop yield prediction, machine learning, South Asia, Mean Squared Error (MSE), R² score

Citation:

S.Thavareesan, Exploring Machine Learning Techniques For Accurate Crop Yield Prediction. Annals of Agricultural Science And Technology 2025.