Replication Package for Explainable AI for Software Engineering

This repository contains the supporting material for the book Explainable AI for Software Engineering: From Prediction to Explanation Reliability and Trust. The material was assembled to make the empirical results reported in the book easier to inspect, trace, and reproduce where the original experimental artifacts are available. The package includes figures used in the book, a registry of experimental results, out-of-fold diagnostic records from the KC1 defect prediction study, and structured CSV versions of selected result tables. These materials cover the main empirical parts of the book, including software defect prediction, SHAP explanation stability and robustness, cross-project explanation analysis, and the God Class case study. Some limitations of the archive are worth noting. The CSV files derived from manuscript tables preserve reported results but are not substitutes for the original raw experimental outputs. Third-party datasets are also not redistributed in this archive. Where possible, provenance and integrity information is provided so that the data used in the analyses can be identified independently. The archive follows the same principle adopted throughout the book: results are included only when their provenance can be stated clearly. Missing historical experimental details have not been reconstructed retrospectively, and experiments for which no traceable numerical record is available are not presented as completed results.