Using public domain metrics to estimate software development effort

In this paper we investigate the accuracy of cost estimates when applying m ost commonly used modeling techniques to a large-scale industrial data set which is professionally maintained by the International Software Standards Benchmarking Group (ISBSG). The modeling techniques applied are ordinary le ast squares regression (OLS), Analogy-based estimation, stepwise ANOVA, CAR T, and robust regression. The questions we address in this study are related to important issues. The first is the appropriate selection of a technique in a given context The s econd is the assessment of the feasibility of using multi-organizational da ta compared to the benefits from company-specific data collection. We compare company-specific models with models based on multi-company data. This is done by using the estimates derived for one company that contribut ed to the ISBSG data set and estimates from using carefully marched data fr om the rest of the ISBSG data. When using the ISBSG data set to derive estimates for the company generally poor results were obtained. Robust regression and OLS performed most accur ately. When using the company's own data as the basis for estimation OLS, a CART-variant, and Analogy performed best. In contrast to previous studies, the estimation accuracy when using the com pany's data is significantly higher than when using the rest of the ISBSG d ata set. Thus, from these results, the company that contributed to the ISBSG data se t, would be better off when using ifs own data for cost estimation.