Date of Award

9-1995

Document Type

Thesis

Degree Name

Master of Science in Cost Analysis

First Advisor

Daniel V. Ferens, PhD

Abstract

Several sophisticated Department of Defense (DOD) weapon and space systems depend on a variety of mission-critical computer software applications. Because of the increasing costs to develop these applications, and today's austere defense budget, accurately estimating the costs to develop software is gaining a great deal of attention within the DOD. Unfortunately, software cost estimation models have been encountering difficulties in accurately estimating the costs of software development. However, research has shown that calibrating one's software cost estimation model can improve the predictive capability of that model. As such, this study focused on calibrating the SLIM cost model to sets of historic software projects that were described along a specific development environment, and assessing the predictive ability of the model resulting from this calibration. The SMC database was first stratified along specific operating environment and application parameters, as well as some key SLIM input parameters (size, effort, development schedule, and minimum peak staffing). Stratification of the SMC database resulted in three data subsets, two of which were defined along a specific operating environment and application type, and one which was only defined along a specific operating environment. Once SLIM was calibrated to each-data set, the calibrated SLIM models were validated in an effort to verify their predictive ability. Validation was accomplished using MRE, MMRE, percentage method, and standard deviation measures.

AFIT Designator

AFIT-GCA-LAS-95S-6

DTIC Accession Number

ADA301603

Comments

The author's Vita page is omitted.

Presented to the Faculty of the Graduate School of Logistics and Acquisition Management of the Air Force Institute of Technology.

Download

Share

COinS