Software Support Cost Estimating Models: A Comparative Study of Model Content and Parameter Sensitivity

摘要

This research entailed a comparison of five software estimating models: PRICE-S, SEER-SEM, SoftCost-OO, SoftEst, and SPR KnowledgePLAN. The objective was to research the differences of the software models as related to software support cost. The following major question areas were addressed: (1) How do the differences between the models impact the resulting cost estimates, (2) To what degree can we explain and adjust for the differences between cost models, All items were for flight avionics of a manned aircraft. The differences between the models significantly impact the resulting estimates. Over the five models evaluated, a range of over $60 million occurred during a twenty year estimate. The researchers can explain the differences in the models due to the different algorithms used, but were not able to normalize the models to achieve equivalent estimates. The researchers feel a typical user will not be able to normalize separate models and should, therefore, concentrate on learning one or two models in detail. Different models are more appropriate depending on the task or project being estimated.