Due to advances in the data mining (DM) field, it is possible to extract knowledge
from raw data. Indeed, powerful techniques such as neural networks (NNs)
and more recently support vector machines (SVMs) are emerging. While being
more flexible models (i.e. no a priori restriction is imposed), the performance depends
on a correct setting of hyperparameters (e.g. SVM kernel parameter) and
the input variables used by the model. In this study, we present an integrated
and computationally efficient approach that simultaneously addresses both issues.
Sensitivity analysis is used to extract knowledge from the NN/SVM models,
given in terms of the effect on the responses when one input is varied, leading
to the proposed Variable Effect Characteristic (VEC) curves, and relative importance
of the inputs (measured by the variance of the response changes). The
the variable selection is guided by sensitivity analysis and the model selection is
based on parsimony search that starts from a reasonable value and is stopped
when the generalization estimate decreases