benzo32 Dataset

This example uses benzo32.arff dataset from the Harrison et al. paper [1].
wGMDH has been run with standard parameters, using optimal network depth to prevent overfitting. Optimal network depth has been predetermined using 10-fold cross-validation.

You can browse through all models found suitable for building more complex ones. The model with best performance, set as wGmdh classifier output, is situated on the left of topmost layer.

[1]   Harrison,P.W. and Barlin,G.B. and Davies,L.P. and Ireland,S.J. and Matyus,P. and Wong,M.G., Syntheses, pharmacological evaluation and molecular modelling of substituted 6-alkoxyimidazo[1,2-b]pyridazines as new ligands for the benzodiazepine receptor, European Journal of Medicinal Chemistry, (31), 1996, 651-662