Doctoral Dissertations

Date of Award

12-2000

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Computer Science

Major Professor

Jesse H. Poore

Committee Members

Michael W. Berry, Kenneth C. Gilbert, Thomas E. Potok, Robert C. Ward

Abstract

Justification of investments to improve software development processes and technol- ogy continues to be a significant challenge for software management. Managers interested in improving quality, cost, and cycle-time of their products have a large set of methods, tools, and techniques from which to choose. The implementation of one or more of these potential improvements can require considerable time and cost. Decision makers must be able to understand the benefits from each proposed improvement and decide which improvements to implement. While a variety of approaches exist for evaluating the costs and benefits of a few specific improvements such as inspections or systematic reuse, there is no general model for evaluating software process improvements.

The result of this research is a practical, useful framework to assist practitioners in evaluating potential process improvements. This general framework can accommodate a variety of methods for estimating the cost-benefit effects of a process change. To illustrate this framework a set of cost-benefit templates for Emerald and Cleanroom technologies were developed and validated. Methods for evaluating effects range from constants and simple equations to bayesian decision models and dynamic process simulations. A prototype tool was developed to assist in performing cost-benefit analysis of software process improvements.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS