The structural characteristics of transportation networks (SCTN) under sustainable development goals are crucial for the development and construction of urban agglomerations (UAs). Evaluating the SCTN in UAs can be difficult due to the increasing diversity of UAs and the multiple interactive criteria involved. To address these challenges, this paper develops the G-DEMATEL model, a new data-driven model that integrates the lambda-step gravity model and the decision-making trial and evaluation laboratory method. This model is then applied to various scales of Chinese UAs. The results demonstrate that: (1) three types of SCTN have developed in UAs, including polycentric, two-centers, and monocentric networks; (2) the transportation networks in UAs exhibit an excess of independent-type cities and a lack of functional differentiation among cities. (3) SCTN are significantly influenced by natural and location conditions such as geography and distance from the central city, and the gravitational effects of provincial capitals are especially significant.