Abstract:
In this paper, we explore the behaviour of neural
network in predicting software readiness. Our neural network
model aims to predict the number of faults (including objectoriented faults) of a software under development. We use Ward
neural network that is a backpropagation network with different
activation functions. Different activation functions are applied to
hidden layer slabs to detect different features in a pattern
processed through a network. In our experiments, hyperbolic
tangent, Gaussian, Gaussian-complement and linear functions are
used as activation functions to improve prediction. This paper
also compares the prediction results from multiple regression
model and neural network model. Object-oriented design metrics
are used as the independent variables in our study. Our study is
conducted on three industrial real-time systems that contain a
number of natural faults that has been reported over a period of
three years.
