WCCI Performance Prediction Challenge

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

cv5_prepro_nonlinear_svc

Submitted by Reference

subsampling 3000 examples , standartization and classification using kernel which is a product of linear and gaussian

BER estimation by 5 fold CV

CLOP definition:
my_svc=svc({'coef0=1', 'degree=1', 'gamma=0.001', 'shrinkage=0.1'});
for sylva subsampling of 3000 points was performed prior to classification
for ada, gina and hiva standartizaion was also performed

Dataset Balanced Error Test guess Guess error Test score Area Under Curve
Train Valid Test Train Valid Test
ada 0.1288 0.2521 0.2401 0.2239 0.0162 0.2563 0.9666 0.8864 0.8738
gina 0 0.054 0.064 0.0734 0.0094 0.0734 1 0.9817 0.9821
hiva 0.0001 0.467 0.4247 0.4446 0.0199 0.4442 1 0.6618 0.7386
nova 0 0.112 0.0972 0.1113 0.0141 0.1113 1 0.9747 0.982
sylva 0.0469 0.0641 0.0578 0.0349 0.0229 0.0807 0.9979 0.9983 0.9964
Overall 0.0352 0.1898 0.1768 0.1776 0.0165 0.1932 0.9929 0.9006 0.9146