Abstract—Estimation of development effort in software projects has been a challenging issue since many years ago. Uncertain behavior of software projects makes the estimating process difficult at early stage of project so that achieving to accurate estimations seems to be impossible in software projects. Neural Networks(NN) and Analogy Based Estimation (ABE) methods have been widely used in this field because the nature of which is adaptable with dynamic environment of software projects. Since most software project datasets include some irrelevant and inconsistent projects, the quality of neural network training and also quality of ABE estimations have been significant problems in all prior studies. In this paper, to overcome the problem of inconsistence projects, fuzzy clustering has been used for placing the similar projects in several clusters. A new framework was proposed to combine ABE and NN using C-Means clustering. The results showed that the proposed method improved the performance of NN and ABE noticeably.

Index Terms—Development effort estimation, inconsistent projects, analogy based estimation, fuzzy clustering, neural network

B. V. Khatibi is with the Bardsir Branch, Islamic Azad University,Kerman, Iran.
D. N. A. Jawawi and E. Khatibi are with the Department of Software Engineering,Faculty of Computer Science and Information System,Universiti Teknologi Malaysia (UTM), 81310 Skudai, Johor, Malaysia(e-mail: dayang.n.a.jawawi@gmail.com).

Cite: Vahid Khatibi. B, Dayang N. A. Jawawi, and Elham Khatibi, "Increasing the Accuracy of Analogy Based Software Development Effort Estimation Using Neural Networks," International Journal of Computer and Communication Engineering vol. 2, no. 1, pp. 78-81 , 2013.