A new approach to software project cost estimation using a hybrid model of radial basis function neural network and genetic algorithm

Abstract

The use of Artificial Neural Networks (ANNs) is growing day by day in various areas of software engineering and the relevant studies are significantly focused on the software project cost estimation. The problem of high accuracy cost estimation is one of the most important issues, which is still a challenge for researchers in the area of software. In this paper, a new method is presented based on Radial Basis Function (RBF) neural network and Genetic Algorithm (GA) to estimate software project cost. So, to provide optimal models, we tried to achieve good results by selecting proper parameter and network architecture. To prove the optimality of our model, the output results of the proposed model were compared with those of the estimation via the COCOMO81 algorithmic model. Test results show that the proposed adaptive models provide greater accuracy than algorithmic models, without the need to involve estimator in the details and complex relationships of the attributes and drivers of software project cost.