Date of Award

9-1995

Document Type

Thesis

Degree Name

Master of Science in Cost Analysis

First Advisor

Daniel V. Ferens, PhD

Abstract

This study examined the change in performance resulting from calibration of the REVIC software cost model and whether calibration improved the model's ability to estimate software development costs. Three calibrations were conducted on two operating environments using the SMC database. The calibrations consisted of one coefficient only and two coefficient and exponent calibrations, using different procedures. Analysis of the goodness of fit" was accomplished using MRE, MMRE, RMS, RRMS, and the prediction test to measure the model's improvement in predicting ability. The Wilcoxon and Wilk-Shapiro tests were used to test for normality of the sample data. The author concluded that the REVIC model's predicting ability is insufficient when more than one independent variable is impacting cost. IR the case of the two operating environments selected, military ground and unmanned space, the improvement in the estimating ability of the model, following calibration, was insufficient to predict cost in either the military ground or the unmanned space environment within an acceptable confidence level.

AFIT Designator

AFIT-GCA-LAS-95S-13

DTIC Accession Number

ADA300694

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