Background: Changes on source code may propagate to distant code entities through various relationships, making related changes obligatory. Identifying the change impacts is challenging due to the intricate ways in which changes spread. Association rules have been widely used in change impact prediction. However, such rules only depend on historical co-changes, limiting the prediction accuracy if entities rarely or never co-changed. Aims: This study explores the integration of evolutionary coupling with software change relationships among changed code entities to enhance the state-of-the-art association rule mining technique, TARMAQ. Method: We integrate evolutionary coupling with 12 types of software change relationships, such as structural dependencies and code clones, to better capture the associated changes.
Results: Analyzing thousands of commits from six open-source systems, we observed: (1) Incorporating software change relationship analysis significantly improves TARMAQ’s prediction recall and mean average precision (MAP), (2) The top-5 predictions exhibit notable increasing in precision, recall, F1-score, and MAP, and (3) Our tests confirm the practical applicability of this integrated impact prediction method. Conclusions: We conclude that combining evolutionary coupling and software change relationships improves the recall and prioritization of impact predictions in association rule-based techniques.