Advancing pearl millet yield forecasting - Comparative analysis of individual and ensemble machine learning approaches over Rajasthan, India

Aug 5, 2026

Upload by Asher

Pearl millet (Pennisetum glaucum L.) is a resilient staple for arid and semi-arid regions, and Rajasthan, India is a leading producer. This study improves pearl millet yield forecasting across nine Rajasthan districts (1997–2019) using weather inputs from NASA POWER and yield data from the Directorate of Economics and Statistics. Individual ML models (GLM, ELNET, XGB, SVR, RF) are compared with ensemble variants. Model selection relies on R2 and nRMSE rankings over training and testing, revealing location-dependent performance and recommending ensemble ELNET, then individual RF, to support accurate prediction under differing geographic and environmental conditions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages24

Reading Time1 h 0 min

AccessFree to download

Abstract

Introduction

Agricultural context and precision agriculture need

Pearl millet significance and policy support

Materials and Methods

Study region and data sources

Model selection and evaluation approach

Results and Discussion

Comparative performance across districts

Training-validation behavior and overfitting considerations

Recommended model strategy

OPEN ACCESS Citation: Alsaber A, Setiya P, Satpathi A, Aljamaan A, Pan J (2025) Advancing pearl millet yield forecasting: Comparative analysis of individual and ensemble machine learning approaches over Rajasthan, India. PLoS ONE 20(3): e0317602 . [https://doi.org/10.1371/](https://doi.org/10.1371/)[ ](https://doi.org/10.1371/)[journal.pone.0317602](journal.pone.0317602) Editor: Bappa Das, ICAR Central Coastal Agricultural Research Institute, INDIA Received: July 30, 2024 Accepted: December 31, 2024 Published: March 11, 2025 Copyright: © 2025 Alsaber et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data availability statement: The data can be accessed through the following public repository link: [https://github.com/alsaberACS/](https://github.com/alsaberACS/)[ ](https://github.com/alsaberACS/)[pearl-millet-data.git](pearl-millet-data.git.)[.](pearl-millet-data.git.) Funding: This publication was made possible by the support of the AUK Open Access Publishing Fund. RESEARCH ARTICLE Advancing pearl millet yield forecasting: Comparative analysis of individual and ensemble machine learning approaches over Rajasthan, India Ahmad Alsaber1*, Parul Setiya2, Anurag Satpathi3, Abrar Aljamaan4, Jiazhu Pan5 1 Department of Management, College of Business and Economics, American University of Kuwait, Salmiya, Kuwait, 2 Govind Ballabh Pant University of Agriculture and Technology, Pantnagar, Uttarakhand, India, 3 Division of Agrometeorology, Sher-e-Kashmir University of Agricultural Sciences and Technology, Kashmir Shalimar campus, Srinagar, India, 4 College of Art and Science, American University of Kuwait, Slamiya, Kuwait, 5 Department of Mathematics and Statistics, University of Strathclyde, Glasgow, United Kingdom * [aalsaber@auk.edu.kw](aalsaber@auk.edu.kw) Abstract Pearl millet (Pennisetum glaucum L.) is a resilient crop known for its ability to thrive in arid and semi-arid regions, making it a crucial staple in regions prone to drought. Rajasthan, a state in India, emerged as the top producer of pearl millet. This study enhances yield forecasting for pearl millet using machine learning models across nine districts viz. Jaipur, Ajmer, Jodhpur, Bikaner, Bharatpur, Alwar, Sikar, Jhunjhunu and Nagaur in Rajasthan, India. Data from 1997–2019 (23 years), including yield data from the Directorate of Economics and Statistics and weather data from the NASA POWER web portal, were analysed. The study employed individual machine learning methods (GLM, ELNET, XGB, SVR and RF) and their ensemble combinations (GLM, ELNET, Cubist and RF) . Discerning the overall best performing model across all locations remained challenging. For instance, while ensemble models exhibited subpar performance in Barmer and Nagaur, their performance ranged from satisfactory to commendable in other locations. To identify the best model, all models were ranked based on their R2 and nRMSE (%) values. Combined average ranks during training and testing revealed the model performance ranking as I-XGB (3.83) > I-GLM (4.28) > EELNET (4 .32) > I-RF (4 . 67) > E-GLM (4 . 88) > I-SVR (4 . 90) > I-ELNET (4 . 94) > ERF (6 .03) > E-Cubist (7. 15), where I denotes individual model, while E denotes ensemble model. Intriguingly, while individual GLM and XGB models demonstrated superior performance during calibration, they exhibited poorer performance during validation, potentially indicating issues of data overfitting. Hence, the ensemble ELNET approach is recommended for accurate prediction of pearl millet yield, followed by the individual RF model. These performances underscore the importance of tailored model selection based on specific geographic and environmental conditions. Competing interests: The authors have declared that no competing interests exist. Intr

FAQs

Which machine learning models are compared for pearl millet yield forecasting?

The study evaluates individual models including GLM, ELNET, XGB, SVR, and RF, and compares them with ensemble combinations such as GLM, ELNET, Cubist, and RF.

What data range and sources are used in the analysis?

Yield data from the Directorate of Economics and Statistics and weather data from the NASA POWER web portal are analyzed for 1997–2019 across nine Rajasthan districts.

How are models ranked to determine the best performers?

Models are ranked using R2 and nRMSE (%) values, using combined average ranks from training and testing to establish performance ordering.

Which modeling approach is recommended for accurate yield prediction?

The ensemble ELNET approach is recommended for accurate pearl millet yield prediction, followed by the individual RF model based on the ranking results and validation behavior.

Aug 5, 2026

Upload by Asher