An evaluation of three function point models for estimation of software effort

It is pointed out that function point analysis has become well-established as a method for estimating software size and, sometimes, software effort. However, the accuracy of estimation models using function points to estimate size or effort has not been ascertained significantly. The authors summarize a study performed to assess the expected accuracy of three function point-based models: the Tecelote Software Program Acquisition Network Simulation (SPANS) model, the Checkpoint model, and the Costar model. It is shown that the three function point models can be useful in effort estimation, especially if they can be adjusted for bias. Based on the percentage tests, the unadjusted Checkpoint model was more accurate than the other two models. However, when models were adjusted for bias, the results were very similar.< >