Abstract:Industrial clusters clustered in different regions show differences in innovation efficiency and innovation capabilities, and different clusters should correspond to different innovation efficiency improvement paths. Based on the data of my country's innovative industrial clusters, first use the data envelopment analysis (DEA)-Malmquist index method to calculate the innovation efficiency of clusters, which can be divided into comprehensive, technologically advanced, technologically weak, and economies of scale. And then use fuzzy set qualitative comparative analysis (fsqca) to identify 7 paths that affect the innovation efficiency of clusters in different situations, and divide them into comprehensive, environmental, and aggregation types, and obtain the innovation improvement paths of different types of clusters. The results show that: (1) The innovative industrial clusters in Beijing are the only ones that are comprehensive, the innovative industrial clusters in Fujian, Guizhou, Chongqing, and Hubei are technologically advanced, and the innovative industrial clusters in Inner Mongolia, Jilin, and Yunnan are technologically weak. The innovative industrial clusters in Jiangsu, Zhejiang, Sichuan, and Tianjin belong to economies of scale; (2) The comprehensive path is suitable for comprehensive and technologically advanced clusters, the technologically weak clusters are suitable for environmental paths, and the aggregation path is suitable for technological progress and economies of scale cluster, but the same type of industrial cluster can correspond to different paths in different regions to achieve the effect of improving innovation efficiency; (3) The number of clusters, the number of industrial alliances and the number of invention patents granted in the year are the main indicators that affect the innovation efficiency of innovative industrial clusters .