Purpose: This study proposes a new Burr type Ⅹ-based NHPP software reliability model (SRM) to describe failure detection behavior under uncertain operating environments and to compare its performance with existing models.Methods: Model parameters were estimated using the least squares estimation method based on two real software failure datasets. Model performance was evaluated using several criteria, including MSE, MAE, MEOP, PC, and adjusted R2, along with a Multi-Criteria Decision Making Ranking (MCDMR) approach.Results: The results indicate that the proposed model provides superior performance for most evaluation criteria across both datasets. The mean value function and 95% confidence interval analysis show that the proposed model adequately represents the observed failure data.Conclusion: The proposed Burr type Ⅹ-based SRM effectively reflects uncertainty in failure detection processes and demonstrates improved predictive performance compared with existing software reliability models.
Song et al. (Tue,) studied this question.