Censored life data analysis in software testing
Mehmet Şahinoglu, C. Vural · 2002
The main idea is to adapt the censored life testing model to a computer software testing problem. Therefore the goal is not only to compute mean line code to failure (MLTF) or mean time to failure (MTTF), but also the reliability of the software product at a given time "t" or "line of code". It must be also noted that all computations to analyze the system are performed by regarding two types of censoring called Type I and Type II. The numerical applications indicate that the censored results approach the true but unknown uncensored (complete) estimators as the degree of censoring is reduced to zero towards the end of completed life test study. Further, a bias correction procedure for the reliability is implemented.>