Reading the way labels are being handled in the training script and pred script, for each label (i.e. "Obesity" and the co-morbidities) the classes simply converted to binary (if intuitive[name] is not None and intuitive[name] == 'Y': label[idx] = 1). More importantly, the evaluation seems to have been conducted under the (multi-label) binary classification setting.
Is this correct or have I missed something important here? If so, why would it be comparable to the results of the original challenge?
Reading the way labels are being handled in the training script and pred script, for each label (i.e. "Obesity" and the co-morbidities) the classes simply converted to binary (
if intuitive[name] is not None and intuitive[name] == 'Y': label[idx] = 1). More importantly, the evaluation seems to have been conducted under the (multi-label) binary classification setting.Is this correct or have I missed something important here? If so, why would it be comparable to the results of the original challenge?