A Preference-Integrated Assurance-Region Multiplicative DEA Framework for Provincial Site Selection of Cassava Collection and Processing Hubs in Northeastern Thailand

Data Envelopment Analysis (DEA) offers flexible endogenous weighting but was originally developed for input–output efficiency assessment rather than complete multicriteria ranking. This study develops a Preference-Integrated Assurance-Region Multiplicative DEA (PI-AR-MDEA) framework that advances DEA toward a stand-alone multi-criteria decision-making (MCDM) method. The framework combines a unified benefit–cost multiplicative structure with alternative-specific endogenous criterion exponents constrained by Assurance Region bounds around an external preference structure. CRITIC is used to derive objective reference weights, while PI-AR-MDEA ranks alternatives directly from positive raw data through unit-invariant pairwise ratios. In the FMS benchmark, setting γ=1 reproduced the published FMDEA ranking exactly, with a maximum score deviation of 0.000069. In the empirical assessment of 20 Northeastern Thai provinces using six criteria, Nakhon Ratchasima, Khon Kaen, and Ubon Ratchathani ranked first to third under all seven compared methods, and PI-AR-MDEA showed the strongest agreement with FMDEA (ρs=0.996992). Across 36 weight-perturbation scenarios and 1,500 Monte Carlo runs, the first-ranked province and ordered top three were retained in 100% of cases. These findings indicate that PI-AR-MDEA provides a preference-constrained, adaptive, and robust extension of DEA for multicriteria ranking and strategic regional screening.