Re-analysis of JHipster failures/bugs with learning/interpretable techniques
https://github.com/xdevroey/jhipster-dataset/raw/master/v3.6.1/test/jhipster3.6.1-testresults.csv
is needed
- The statistical learning part is not working (TODO)
- The failure analysis is there, though some inconsistencies wrt EMSE results (Table 2 and text) remain
- Key result: there are configurations with 2 or 3 bugs (but only one failure pops out) where a bug is a combination of features (feature interactions?) necessarily leading to a build failure