When the traditional single evaluation index method faces multiple inputs and outputs, it becomes a problem to assign appropriate weights to each input and output. The DEA method based on linear regression mathematical modeling evaluates the efficiency of each decision unit by comparing their relative efficiencies in the absence of an explicit production function or efficiency criterion. The method can consider multiple input and output factors simultaneously and automatically assigns weights to them to evaluate the performance and efficiency of the same or different units in a more comprehensive and objective way. In this study, static DEA and dynamic DEA are used to measure the innovation efficiency of 95 Chinese listed chemical drug manufacturers in the last 5 years. On this basis, the firms were categorized in quadrant diagrams after unidimensional processing of input and output indicators using principal component analysis.