A number of software effort estimations have attempted using statistical mo
dels, case based reasoning, and neural networks. The research results showe
d that the neural network models perform at least as well as the other appr
oaches, so we selected the neural network model as the estimator. However,
since the computing environment changes so rapidly in terms of programming
languages, development tools, and methodologies, it is very difficult to ma
intain the performance of estimation models for the new breed of projects.
Therefore, we propose a search method that finds the right level of relevan
t cases for the neural network model. For the selected case set, the scale
of the neural network model can be reduced by eliminating the qualitative i
nput factors with the same values. Since there exist a multitude of combina
tions of case sets, we need to search for the optimal reduced neural networ
k model and corresponding case set. To find the quasi-optimal model from th
e hierarchy of reduced neural network models, we adopted the beam search te
chnique and devised the case-set selection algorithm. We have shown that th
e resulting model significantly outperforms the original full model for the
software effort estimation. This approach can be also used for building an
y case-selective neural network. (C) 2001 Elsevier Science Ltd. All rights
reserved.