City Research Online

Abstract

This thesis examines the problem of software reliability growth: how to measure it and how to know that the measures (predictions) are accurate.
Models in continuous time (i.e. complete inter-event time data) and discrete time (i.e. counts of numbers of events in successive observation periods) are considered. Several of these models are new, or are new versions of existing models.

Several statistical tools are presented which will allow a user of the models to analyse their respective merits (predictive accuracy) on a particular data set. Examples of the use of these tools on the predictions from several models on several real data sets are presented. The models perform with variable accuracy, which suggests that no model can be trusted to be of universal usefulness. The techniques presented here, then, form the beginnings of a tool-set which will enable a user to obtain reliability predictions for a particular context, and know that they are accurate.

Publication Type:	Thesis (Doctoral)
Subjects:	Q Science > QA Mathematics Q Science > QA Mathematics > QA76 Computer software
Departments:	School of Science & Technology > Mathematics School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads