Bayesian Software Prediction Models. Volume IV. Bayesian Software Correction Limit Policies.

摘要

This report deals with the problem of determining an optimum correction limit policy for a large software system subject to random occurrences of errors in an operational phase. When an error occurs, the corrective action is scheduled for either the programmer (Phase I) or the system analyst (Phase II). Two cost models are developed and procedures for obtaining the optimum correction time T* are described to minimize the expected cost per unit time. Numerical examples are given to illustrate the procedure. (Author)