RECOMMENDATIONS FOR IMPROVING SOFTWARE COST ESTIMATION IN DOD ACQUISITION

Abstract

Acquisition initiatives within the Department of Defense (DOD) are becoming increasingly reliant on software. While the DOD has ample experience in estimating costs of hardware acquisition, expertise in estimating software acquisition costs is lacking. The objective of this capstone project is to summarize the current software cost estimating methods, analyze existing software cost estimating models, and suggest areas and methods for improvement. To accomplish this, surveys were conducted to gather program cost data, which was run through existing cost estimating models. From here, the outputs were compared to actual program costs. This established a baseline for the effectiveness of existing methods and guided suggestions for areas of improvement. The Software Resource Data Reports (SRDR) data used seemed to have spurious data reporting from at least one source, and the base cost estimation models were not found to be sufficiently accurate in our study. The capstone finds that calibrating the cost models to the data available improved those models dramatically. In all, the capstone recommends performing data realism checks upon SRDR submissions to ensure data accuracy and calibrating cost models for each contractor with the available data before using them to estimate DOD Acquisition costs.