Metrics for Software Reliability Failure‐Count Models in Cyber‐Risk
Mehmet Şahinoglu · 2016
The conventional nonhomogeneous Poisson (NHPP) models do not treat or permit the multiple counts as the compound Poisson∩geometric model and compound Poisson∩LSD (or negative binomial) properly execute. These two models are superior to other competing models such as Musa–Okumoto's (M–O) logarithmic Poisson method—already studied in Chapter —in most data sets, if not all, with clumping of failures in discrete effort domain. CPM is a technique to predict the failure count at a future time. It clearly takes into account the clumping effect in an NHPP process by using the compound Poisson, where clustering was not modeled earlier in calculations. The probability density estimation of the number of software failures in the event of clustering or clumping of the software failures is the subject of this section in this chapter. Absolute relative error (RE) and squared relative error (SqRE) are random variables suggested as measurements of the forecast accuracy of the total number of software failures estimated at the end of a mission time. The purpose is to compare the predictive merit of competing software reliability models, an important metrics-based concern to software reliability analysts. This technique calculates the Bayes probability of how much better the prediction accuracy is for one method relative to competitors. The same idea can be applied to security risk-based failure counts.