Abstract: Based on a reverse gravity model, endogenous nodal attractions of 28 cities are estimated using the algebraic method with a Chinese railway passenger database with known exogenous spatial interaction and impedance. Then nodal attributes that contribute to the nodal attractions are examined using correlation index and linear regression function. We find that the 28 cities'nodal attractions show a decreasing trend from east to west and the hierarchical structure presents a "spindle" feature. Among many factors, the employees is the best indicator of nodal attractions and whether the city is a transport hub or not has an important influence on its nodal attraction in the "axis - spoke" network. The linear fitting results of nodal attractions indicate that the composite index can achieve better fitting effect than a single indicator.

Key words: nodal attraction, railway passenger flow O-D matrix, reverse gravity mode, the algebraic method