A three- stage algorithm for software cost and time estimation in fuzzy environment

Abstract

The software productions are very expensive projects with uncertainty. For these reasons, the estimation of time and cost of software isvery important for both producers and consumers. In this paper an efficient threestage algorithm is developed for software production cost and time estimation. In the first stage, the required personmonth for implementation of software production are obtained by COCOMO and function point methods. We integrated these two methods to consider all aspects in software production and increase estimation accuracy. In the second stage the required duration for completion of each step of production (planning, analysis, design and so on) is obtained by paired comparisons matrix. In third stage, tables of complete time and cost of software are concluded by GERT network in project control and work break structure (WBS). In whole of all stages of this paper, triangular fuzzy numbers are used to express uncertainty existed in succession and repetition of each production step, time of beginning, ending, the duration of each task and costs of them. Retrieved results examined by 30 practical projects and conclude accuracy of 93 percent for time estimation and 92 percent for cost one. Also suggested algorithm is more accurate than COCOMOІІ 2000 algorithm as 50 percent based on examined problems.