Function Point Tree-Based Function Point Analysis: Improving Reproducibility Whilst Maintaining Accuracy in Function Point Counting

We propose a method to improve reproducibility whilst keeping accuracy for the Function Point Analysis (FPA) method. The proposed method is based on a new artifact model called Function Point Tree (FPT). FPT enables a standardized and systematic collection of all data required for FP counting. The new measurement method is called Function Point Tree-based Function Point Analysis (FPT-FPA). We designed FPT-FPA to comply with the IFPUG's FPA steps. We implemented a prototype tool to show the feasibility of automation of the proposed method as well as to support its evaluation. We conducted an empirical study to evaluate FPT-FPA. Our results show general coefficients of variation lower than the maximum expected for both reproducibility and accuracy when compared to the standard FPA method.