Post-Verification of an Optimization Algorithm for Military Food Waste Prediction and Requisition
Journal of the Korea Management Engineers Society · 30.06.2026 · 31 (2) · 149-161
DOI: 10.35373/kmes.31.2.10 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of the Korea Management Engineers Society, online erschienen am 30.06.2026. Verfasst von Jae-Woo Lee und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Purpose
This study validates a military food service optimization algorithm, evaluating its impact on plate waste, procurement claims, and model generalization.
Methods
Using 8,983 records, pre/post-implementation changes were evaluated via Welch tests, HC3-robust OLS, and DID. Additionally, eleven machine-learning models were trained on pre-implementation data and externally validated using post-implementation results.
Results
Per-capita plate waste dropped by 72.4% per meal and 73.8% per day. DID estimates revealed claim reductions of 651.6 KRW (overall) and 317.1 KRW (meat) at the 10% significance level. Externally, Linear Regression, XGBoost, CatBoost, and ensembles surpassed the random forest baseline under distributional shift.
Conclusion
The algorithm significantly reduced plate waste and procurement claims, indicating that model selection should prioritize generalization and periodic retraining over in-sample performance.
Zitieren
APA
Lee, J., Kim, H., & Kim, H. (2026). Post-Verification of an Optimization Algorithm for Military Food Waste Prediction and Requisition. Journal of the Korea Management Engineers Society, 31(2), 149-161. https://doi.org/10.35373/kmes.31.2.10
Vancouver
Lee J, Kim H, Kim H. Post-Verification of an Optimization Algorithm for Military Food Waste Prediction and Requisition. Journal of the Korea Management Engineers Society. 2026;31(2):149-161. doi:10.35373/kmes.31.2.10
BibTeX
@article{Lee2026,
author = {Lee, Jae-Woo and Kim, Hyeon-Jun and Kim, Hyun-Ho},
title = {Post-Verification of an Optimization Algorithm for Military Food Waste Prediction and Requisition},
journal = {Journal of the Korea Management Engineers Society},
year = {2026},
volume = {31},
number = {2},
pages = {149--161},
doi = {10.35373/kmes.31.2.10}
}
EUVIMED