AKWI-Tagungsband zur 35. AKWI-Jahrestagung. Jahrgang, 2022, Seite S. 287–303
Systematic testing of every single component and interface is undoubtedly an important measure to handle the complex nature of current software systems. However, this comes with often neglected computational costs. The aim of this paper is therefore to cut time and resource needs by predictive testing, i.e., predicting test failures with machine learning using a surprisingly simple statistical feature representation. Furthermore, we present the first open research benchmark for pre- dictive testing to enable and foster future research in this area
