Using ZSG-DEA and inverse DEA to predict resource allocation for road traffic safety in China

Abstract

The issue of road traffic safety in China has consistently been a focal point of research within the academic community. Previous literature has proposed the use of Data Envelopment Analysis (DEA) models to evaluate safety efficiency. However, few scholars have examined the associated resource allocation issues or predicted future resource distribution. Consequently, this study aims to forecast resource allocation for future road traffic safety. We utilize road traffic safety indicators from 2012 to 2022 to calculate the resource allocation for 2025 across three aspects. Firstly, we employ the zero-sum gains Data Envelopment Analysis (ZSG-DEA) model in conjunction with the Gray Model (GM(1,1)) and the Criteria Importance Though Inter-criteria Correlation (CRITIC) method to predict the three factors influencing road traffic safety. Secondly, we utilize the inverse DEA model of frontier changes to forecast the necessary investments aligned with the planning objectives. Finally, we apply machine learning methods to predict output. The results indicate that: (1) In certain provinces, the quota for the three factors influencing road traffic safety is limited, resulting in increased pressure for future development. (2) Under the established planning objectives, some provinces require substantial investment, which should be judiciously managed. (3) Certain provinces exhibit high predicted output but must identify the underlying reasons for their low efficiency to enhance overall output. In summary, our research provides data-driven support and recommendations for decision-makers in establishing future planning goals.