An Analysis Of Recurrent Software Errors In A Real-Time Control System

This paper presents an analysis of actual software error data from a large-scale software project using the imperfect debugging model proposed by Goel and Okumoto [1]. The model parameters are estimated from the actual data and the values predicted from the model are compared with the observed values. Joint confidence regions for the parameters are also constructed which permit a study of the sensitivity of predictions.