Fuzzy-GA Based Function Point Model

Accurate cost estimation is the key to software project pricing and the success of resource allocation. Software size is the most important factor in software cost estimation. Among various approaches to software size estimation, FPA (Function Point Analysis )proposed by IFPUG(International Function Point Users Group), is the most widely used. Since the advance of FPA, there have been great changes in many factors related with FPA, including software development approach, database technology and software development language. Therefore, it is of great necessity to do further research on the complexity weight values of FPA. In addition, due to the consideration of convenient calculation and rapid promotion, the traditional way of the metric ranking of complexity weight values fails to explain at length the issue of boundary of complexity categorization and may cause some problems such as crisp boundary and ambiguous complexity classification when calculating project function points.To further clarify the metric ranking of complexity weight values of IFPUG function point model and to make IFPUG function point model more suitable to software size metrics under the conditions of new technology, the thesis researches on the issue of the boundary of complexity categorization and the issue of the effectiveness of complexity weight values of IFPUG FPA. It is proposed in the thesis that an approach based on fuzzy and genetic algorithm, that is, F-Ga-FPA model, can be used to define complexity categorization boundary and calibrate complexity weight values. The main contents of the thesis include: (1) using fuzzy logic theory to extend IFPUG FPA complexity weight classification system from discrete to continuous; (2) using statistical regression methods, combined with the latest database from Interdational Software Benchmark Standard Group (ISBSG) to extract new effort estimation equation; (3) using genetic algorithms, ISBSG and extracted new effort estimation equation to calibration IFPUG FPA; (4) achieving F-Ga-FPA model and applying it to the effort estimation of City Intelligence Parking Guidance System.The results show that reusing the information in function point project warehouse to calibrate complexity weight values of function poing can improve the accuracy of software cost estimation. The improved FPA method performances better than traditional IFPUG FPA during the metrics of City Intelligence Parking Guidance Software.