Jin-Cherng Lin , Han-Yuan Tzeng, and Yueh-Ting Lin, Tatung University, Taiwan

In the IT industry, precisely estimate the effort of each software project the development cost and schedule are count for much to the software company. So precisely estimation of man power seems to be getting more important. In the past time, the IT companies estimate the work effort of man power by human experts, using statistics method. However, the outcomes are always unsatisfying the management level. Recently it becomes an interesting topic if computing intelligence techniques can do better in this field. This research uses some computing intelligence techniques, such as Pearson product-moment correlation coefficient method and one-way ANOVA method to select key factors, and K-Means clustering algorithm to do project clustering, to estimate the software project effort. The experimental result show that using computing intelligence techniques to estimate the software project effort can get more precise and more effective estimation than using traditional human experts did.

Software Effort Estimation, Project Clustering, Computing Intelligence, Particle Swarm Optimization, K-Means Clustering.