Software effort prediction using fuzzy clustering and functional link artificial neural networks

Abstract

We use the combined fuzzy C-Means (FCM) clustering algorithm and functional link artificial neural networks (FLANN) to achieve accurate software effort prediction. FLANN is a computationally efficient nonlinear network and is capable for complex nonlinear mapping between its input and output pattern space. The nonlinearity is introduced into the FLANN by passing the input pattern through a functional expansion unit. The proposed method uses three real time datasets. The Chebyshev polynomial has been used as choice of expansion to exhaustively study the performance. The simulation results show that it not only deals efficiently with noisy data but also proves to be a champion in producing promising results. © 2012 Springer-Verlag.