To scientifically measure and accurately evaluate the spatial quality within the station area and realize effective optimization, seventy-three subway stations in Chengdu City were selected to support multi-source big data such as street network, POI(point of interest), street view pictures, etc., then machine learning and spatial design network analysis(sDNA) and other technologies were used to construct an evaluation system with convenience, functionality and comfort as the core. Large-scale quantitative evaluation of street space quality within the station area was carried out, and guidance and control strategies for different levels of stations were proposed. The results show that 68.03% of the station area streets score is lower than the medium level, the street function and comfort are generally good, and the convenience is poor; at the station level, the street space quality shows the distribution characteristics of high in the South and low in the north, high in the West and low in the East, and high in the inside and low in the outside. The proposed method takes into account the analysis accuracy of human-oriented scale, the analysis depth of site scale and the analysis breadth of urban scale, which is helpful to create an efficient dynamic feedback mechanism of urban management.