This paper first combines the traditional Adaboost iterative algorithm and logistic regression algorithm to construct an improved Adaboost based regression algorithm. In order to solve the problem of the redundant amount or insufficient amount of output of industrial carbon emissions, the SBM model is divided into two stages, and by merging this method, the industrial carbon emission efficiency measuring model is created, While the Global Moran’s I index is used to assess the geographical impact of industrial carbon emission efficiency. Additionally, a model of the influence of emission efficiency based on the geographical effect is built through the selection of the explanatory variables of the influencing factors. According to the study, the industrial carbon emission efficiency is growing at an annual rate of 1.8% during the period of fast expansion, 0.4% in the steady growth stage, and the Z value of STI is 0.38 is significant in spatial autocorrelation.