cytomining / profiling-handbook

Image-based Profiling Handbook
https://cytomining.github.io/profiling-handbook/
Creative Commons Zero v1.0 Universal
8 stars 7 forks source link

Add preselect justifications #41

Closed bethac07 closed 2 years ago

bethac07 commented 5 years ago

In preselect, add justifications for why we use certain sections of the data to do different steps.

shntnu commented 3 years ago

Quick notes

  1. replicate_correlation picks informative features
  2. correlation_threshold drops redundant features
  3. variance_threshold drops features that don't vary

So it's possible for total junk feature to survive the last two, but it won't make it past the first.