WCCI Performance Prediction Challenge

The challenge is over, but a new challenge is on-going using the same datasets, check it out!

cv5+Prepro+linearSVC

Submitted by Reference

This method is based on one of CLOP model examples. It uses standardization as a preprocessing, followed by a linear support vector classifier. estimation of BER is performed via 5 fold cross validation (see below matlab code). the model itself is:
chain({standardize, svc('shrinkage=0.1')})
For nova, no preprocessing is performed.


mfold = 5;
cv_train_output = train(cv(my_model,['folds=' num2str(mfold)]), D.train);

Dataset Balanced Error Test guess Guess error Test score Area Under Curve
Train Valid Test Train Valid Test
ada 0.2204 0.217 0.2384 0.2233 0.0151 0.2535 0.8616 0.8612 0.8476
gina 0.1306 0.1277 0.1406 0.1503 0.0097 0.1503 0.9466 0.9371 0.9347
hiva 0.2488 0.3971 0.3162 0.3169 0.0007 0.3163 0.8677 0.6624 0.7443
nova 0 0.112 0.0966 0.1113 0.0148 0.1113 1 0.9742 0.9816
sylva 0.0859 0.0835 0.0847 0.0842 0.0005 0.0851 0.9978 0.9981 0.9973
Overall 0.1372 0.1875 0.1753 0.1772 0.0082 0.1833 0.9347 0.8866 0.9011