Validation results different after master branch update

Stanford-STAGES / stanford-stages

Automated sleep staging scoring and narcolepsy identification

76 stars 26 forks source link

Validation results different after master branch update #16

Closed informaton closed 4 years ago

informaton commented 4 years ago

On behalf of Wen:

I downloaded the new master version of https://github.com/Stanford-STAGES/stanford-stages. Everything ran smoothly except that I got different results when running inf_narco_app.py on the test file CHP040.edf for both 1-model and 16-model, see below. I wonder if you checked recently.

1-model: Score: -0.4442 (vs -0.0076) Diagnosis: Other

16-model: Score: -0.0287 (vs 0.1658) Diagnosis: Narcolepsy type 1

informaton commented 4 years ago

Investigating.

informaton commented 4 years ago

This turned out to be caused by a switch in one of the narcolepsy feature prediction selection indices (14 for 299) which was updated in the beta branch, which is under development still, but should not have been in the master branch. The indices has been corrected back to their original value and the results verified to match the readme.md file once more.