A robust asymptotic estimate for software error intensity
Show‐Long Patrick Koh · Journal of the Chinese Institute of Engineers · 2009
This paper extends the model of Ross (1985) to include delayed removals and proves that the semi‐parametric, asymptotically unbiased estimate of Koh (1994) is robust by this inefficiency of fault removing. The robustness is consistent with intuitions and is empirically confirmed by trials on a set of real data. The result also indicates that the S‐shaped error intensity, largely due to the early stage of testing period, probably is not an important factor for the asymptotic estimation of the error intensity.