An extensive study of class-level and method-level test case selection for continuous integration.

•A selection framework Sapient can select a precise test subset to reduce test cost.•FEST adopting Sapient can model the full dependency relations at the class level.•MEST adopting Sapient can capture the precise dependencies at the method level.•Promising results for the effectiveness of FEST and MEST, and their comparison.•Public-accessible dataset for the replication and evaluation of other approaches.