Citation

Singla, Sanjay and Senthilpari, Chinnaiyan and Lee, Chu Liang and Tanvir, Zaka Al Shahariar (2026) Crop Yield Prediction using Computationally Efficient Machine Learning Models. Social Science Forum, 10 (8). pp. 850-856. ISSN 03523608

Text 22.pdf - Published Version Restricted to Repository staff only Download (193kB)

Abstract

Within the domain of agricultural precision farming, dependable crop yield forecasting constitutes a fundamental requirement for augmenting productivity, optimizing resource allocation, and facilitating evidence-based decision-making processes. However, the development of a comprehensive representation that adequately captures the multifaceted interactions among climatic, environmental, and agronomic variables remains a considerable obstacle. This investigation presents a crop yield prediction framework utilizing a TensorFlow-implemented Deep Neural Network (DNN) applied to the Large Scale Agriculture Dataset (LSAD), encompassing 19,689 observations across 30 Indian states, 55 crop varieties, and multiple cultivation cycles. To ensure rigorous and impartial model assessment, standardized data preprocessing, hyperparameter tuning, and validation protocols were implemented for comparative analysis with four established machine learning techniques, namely CatBoost, LightGBM, XGBoost, and Random Forest. The empirical findings demonstrate that the proposed DNN delivers superior predictive capacity, attaining the minimum RMSE of 103.89 and the maximum R2 value of 0.9873, thereby demonstrating robust generalization performance across test datasets. Notwithstanding the elevated computational and inference expenses associated with CatBoost, the proposed DNN consistently produces more precise yield estimations. Moreover, the most influential feature ensemble identified through SHAP-based interpretability analysis comprises cultivated area, precipitation volume, temperature, fertilizer application rate, and pesticide application rate, with stability validation confirming the consistency of this feature configuration across repeated analyses. These findings substantiate that the proposed model constitutes a dependable and precise instrument for automated crop yield forecasting and holds considerable applicability in precision agriculture and agricultural policy formulation contexts.

Item Type:	Article
Uncontrolled Keywords:	Deep Neural Network, Explainable artificial intelligence (XAI)
Subjects:	S Agriculture > S Agriculture (General) T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7800-8360 Electronics > TK7885-7895 Computer engineering. Computer hardware
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	05 Oct 2026 04:31
Last Modified:	05 Oct 2026 04:43
URII:	http://shdl.mmu.edu.my/id/eprint/16874

Downloads

Edit (login required)