The identification of source code file experts is an important task related to software development since it can be used to improve software maintenance, code review, bug fixes, among others. In this context, it is necessary to create an effective technique for the automatic identification of these experts. There are studies that propose repository-mining techniques to help identify file experts, but there are still gaps in this area that can be explored. In previous research, it was identified that information such as the recency of modifications and the file’s size influence the knowledge that developers have on the file. In this work, we analyze the relation of these and other variables that can be extracted from the development history of a repository with developers’ source code knowledge. From this analysis, we decided to use machine learning classifiers to guide the identification of file experts. We compare the performances of these classifiers with state-of-the-art techniques, using data from public and private projects. The machine learning classifiers reached the best F-Measure ranging from 71% to 73%, and best Precision ranging from 71% to 73% in both types of projects. We hope the findings of this study will provide insights for future research on building more accurate models.