Enhancing predictive performance in statistical modeling: Innovative hybrid best subset feature selection for rice production in Malaysia
Recent statistics from the Food and Agriculture Organization (FAO) of the United Nations revealed an increasing trend in both moderate and severe food insecurity prevalence in Malaysia on average. To effectively address this issue, comprehensive solutions are needed to consider the four dimensions o...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | en |
| Published: |
Faculty of Science and Technology, UKM
2025
|
| Subjects: | |
| Online Access: | https://umpir.ump.edu.my/id/eprint/46881/1/JQMA2025_1.pdf https://doi.org/10.17576/jqma.2104.2025.04 https://umpir.ump.edu.my/id/eprint/46881/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Recent statistics from the Food and Agriculture Organization (FAO) of the United Nations revealed an increasing trend in both moderate and severe food insecurity prevalence in Malaysia on average. To effectively address this issue, comprehensive solutions are needed to consider the four dimensions of food security, particularly for predicting rice production, a staple food for Malaysians. This study aimed to propose an innovative hybrid deterministic best subset feature selection method for identifying significant determinants impacting rice production in Malaysia, thereby contributing to a more effective understanding and management of food security. These selected determinants aligned with the four dimensions of food security and the key pillars of the Sustainable Development Goals (SDGs). The proposed feature selection method integrates mathematics techniques, specifically the modified Taguchi-based VIseKriterijumska Optimizacija I Kompromisno Resenje (VIKOR) multicriteria decision-making (MCDM) algorithm and three performance metrics. The analysis demonstrated that the proposed method outperformed existing hybrid deterministic feature selection methods in the literature, which lacked a comprehensive consideration of all dimensions of food security. Furthermore, the analysis revealed that the proposed methods consistently achieved higher accuracy compared to automated deterministic wrapper feature selection methods. This article has principally contributed to both the mathematics academia and industry realms. This study provided valuable insight for academicians, policymakers, smallholder farmers, and society by offering a more effective feature selection method, thereby enhancing policy development and farming practices in the context of food security. It contributed significantly to both academic and industry realms by presenting a hybrid deterministic features selection method that enhanced communication and practical application compared to the stochastic metaheuristic features selection method. |
|---|