Automated tests play a crucial role in software development as they provide valuable feedback to developers regarding potential faults. However, as software projects grow in complexity and size, the execution time of test suites can become a significant challenge. In order to address this issue, techniques such as test impact analysis and predictive test selection have been introduced. Both methods aim to accelerate the feedback cycle and optimize the regression testing task by selectively choosing tests based on recent code changes. This results in an improved testing process, ensuring quicker bug detection and more efficient software development. In this thesis, our goal is to conduct a case study on both techniques, in order to compare their efficiency. Test Impact Analysis (TIA) relies on testwise coverage to select and prioritize test cases, while Predictive Test Selection (PTS) employs machine learning algorithms. The techniques were evaluated on two software systems, not only by using real error data, but also by creating new code mutations and inserting new bugs. The results show that by using TIA, we get a shorter time in which the first error is found, compared to using PTS. Furthermore, TIA stands out when it comes to potential time savings. On the other hand, PTS has a remarkable reliability. In fact, it worked for all types of bugs and could correctly select failing tests. Whereas in many cases, TIA was not able to detect bugs due to an inability to collect testwise coverage for some files. Regarding bugs related to code for which TIA knows its testwise coverage, it could achieve a recall of 100%.