Abstract:
Over the past several years, we have developed techniques to discriminate between fault-prone software modules and those that are not, to estimate a software system's residual fault content, to identify those portions of a software system having the highest estimated number of faults, and to estimate the effects of requirements changes on software quality.