Software Sizing for Cost/Schedule Estimation

Abstract

This research thesis presents an investigation into the use of software sizing for cost/schedule estimation, with the assistance of a commercial software development organisation. The investigation was divided into three phases, and involved the use of software sizing with documentation for eleven commercial projects.

Research Phase 1 investigated the value of using different levels of sizing rigour by using the Indicative, Estimated and Full NESMA methods with both early bid stage documentation and later stage functional specifications. The Estimated NESMA method was found to provide the optimum value, in terms of estimation accuracy and effort, but it was only deemed acceptable at the bid stage.

Research Phase 2 investigated how the use of more detailed functional size data could enhance bid stage estimation. A classification of functional categories was developed and used to compare size estimates developed at both the bid and functional specifications stages. The conclusion of this phase was that the more detailed size data is needed to provide a measure of the completeness of bid stage documentation.

Research Phase 3 investigated how the NESMA sizing methods could be adapted to provide greater value. A simplified method was developed and refined using the sample projects, and evaluated against the NESMA methods. The results of this phase showed that the simplified method provided similar estimation accuracy to the Estimated NESMA method, but with a statistically significant relative reduction in the estimation effort.

The main contributions demonstrated by this research were:
• Size estimation should focus on business value rather than estimation accuracy if it is to encourage adoption within industry.
• The level of sizing detail provided by the estimate may be pivotal in determining its value.
• Simplified sizing methods should focus on approximating the level of detail provided by full sizing methods rather than on just the overall functional size.

Date of Award	Nov 2018
Original language	English
Supervisor	Ian Mc Chesney (Supervisor) & FG Wilkie (Supervisor)