The AC nanopore method can measure the impedance of single nanoparticles to obtain information on their material properties as well as their size. One of the technical challenges in applying this capability to bacterial sensing lies in the realization of an analytical method to identify bacteria from measured values. In this study, we improved the bacteria identification performance of the AC nanopore method by using machine learning. Comparing four representative machine learning methods for the classification of bacterial groups that are nearly identical in size and difficult to classify based on size, we found that the random forest method has the best classification performance, achieving a classification accuracy of 78.6% for six different particles containing five bacterial species. The use of machine learning was demonstrated to be effective in improving the performance of the bacterial classification capability in the AC nanopore method.