Change-point Detection for Accuracy Improvement of Software Reliability Assessment  
Author Yuka Minamino


Co-Author(s) Shin Danjo; Masashi Kuwano; Taku Moriyama; Shinji Inoue


Abstract In the software development testing phase, the devel- opment manager may intentionally increase or decrease the number of test personnel or change the fault target according to the testing progress. These changes in the testing environment cause changes in the trend of software reliability growth process. Furthermore, changes in the testing environment are one factor that reduces the accuracy of software reliability assessments based on the existing software reliability growth models (SRGMs). Previous studies have proposed several SRGMs that consider a change-point, with many examples of its application to fault-counting data without allowing for the change-point effect; however, the statistical change-point in the data itself is not considered, including its detection methods. This study uses Change Finder, one of the anomaly detection methods, to detect a statistical change-point from the fault-counting data. We apply the detection result to the delayed S-shaped (DSS) SRGM considering a change-point. The effectiveness of "Change Finder" is confirmed by comparing the goodness-of-fit with a case that considers a change in the testing environment.


Keywords Software Reliability Growth Model (SRGM), Change-point, Change-point Detection, Change Finder
    Article #:  RQD28-279

Proceedings of 28th ISSAT International Conference on Reliability & Quality in Design
August 3-5, 2023