search

AlineTalhouk / splendid

Supervised Learning Ensemble for Diagnostic Identification
https://alinetalhouk.github.io/splendid/
Other
1 stars 0 forks source link

pam error #26

Closed AlineTalhouk closed 7 years ago

AlineTalhouk commented 7 years ago

I just pushed (into the errors data folder) data and a script that results in an error when using PAM this doesn't happen with all datasets or all random seeds only some.

dchiu911 commented 7 years ago

I don't see any new commits yet

AlineTalhouk commented 7 years ago

forgot to sync

dchiu911 commented 7 years ago

Please try again. The issue was this:

When we take the raw predicted classes and filter for only the ones that were classified above the maximum class probability threshold, the resulting predicted class may not span all classes (e.g. c(1, 1, 1, 4, 1, 4, 1) but no 2 or 3). This indicates that there were no "confident" predictions for class 2 or 3. For this scenario, I currently choose to use the original raw predicted classes.