Forecasting time-between-failures of software using fuzzy time series approach

The problem with many of the software reliability models is that they cannot deal with the situations where the time-between-failures information during software testing phase is associated with uncertainty due to fuzzy nature of software testing and debugging. Therefore, in this paper, a procedure is proposed to forecast times-between-failures of software during its testing phase by employing fuzzy time series approach, where time-between-failures of software is represented by a fuzzy set having trapezoidal membership function. The forecasting capability of the proposed approach is examined using software failure data of two real software development projects.