Supply Chain Efficiency Evaluation and Informatization-Driven Mechanism Based on a Multi-Model Integrated Framework

Abstract

This paper constructs an analytical framework based on multi-model fusion for the evaluation and optimization of supply chain operational efficiency. First, multidimensional input and output data undergo extreme value normalization to eliminate dimensional differences and enhance the stability of results. Building on this foundation, the Data Envelopment Analysis (DEA) model is introduced to measure the relative efficiency of each link as an independent decision-making unit, thereby enabling an overall performance evaluation under multi-input–multi-output conditions. Furthermore, the efficiency results are transformed into binary variables and combined with a logistic regression model to characterize the probability of key factors influencing high-efficiency states, thereby revealing the marginal mechanisms of these variables. To examine dynamic changes over time, a fixed-effects regression model is constructed to control for unobservable individual differences, enhancing the interpretability and robustness of the results. Empirical results indicate that IT investment exhibits a consistent positive effect across different models and demonstrates higher sensitivity in the processing stage. The proposed method possesses good structural generality and can be applied to efficiency evaluation and the identification of driving factors across multiple scenarios, providing an extensible modeling approach for the collaborative analysis of multidimensional data in complex systems.