Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

ROI Analysis missing columns after update. #378

Open
hdsliu opened this issue Aug 5, 2024 · 7 comments
Open

ROI Analysis missing columns after update. #378

hdsliu opened this issue Aug 5, 2024 · 7 comments

Comments

@hdsliu
Copy link

hdsliu commented Aug 5, 2024

Describe the bug
I recently updated SIMBA using pip install simba-uw-tf-dev --upgrade. Now when I try to use "Append ROI Features: By Body-Parts", two columns are missing from the feature-extracted csv. They are the "zone_cumulative_time", and "zone_cumulative_percent".
Now the models I have trained previously are no longer usable.

Desktop (please complete the following information):

  • OS: [Windows 11]
  • Python Version [e.g. 3.10]
  • Are you using anaconda? YES
@sronilsson
Copy link
Collaborator

Thanks @hdsliu let me check on this - and make sure that they are there, I will get back to you.

@sronilsson
Copy link
Collaborator

... PS. Do you know which version of SimBA you where using previously, before the update?

@hdsliu
Copy link
Author

hdsliu commented Aug 6, 2024

@sronilsson I am not sure which version was installed. I followed the instruction to use Anaconda to install, earlier this year (March).
I did notice that the GUI has changed (icon and font changes) after the update. So the previous version was before said changes.

@sronilsson
Copy link
Collaborator

Alright, let me look back and include it again. I think the reason I excluded it was (i) I didn't think anyone was using it (not so smart), and (ii) they are not very good features that could bias the model.

@hdsliu
Copy link
Author

hdsliu commented Aug 6, 2024

@sronilsson No need to add them back. I was mostly wondering if it was a random and unpredictable bug. But obviously you removed them for a good reason. I can retrain my models.

@sronilsson
Copy link
Collaborator

What happened was, that someone was using them for training a classifier for different running-styles in a mouse running-wheel. In the annotations for some "fast running" or similar, they were always annotated in the beginning of the session.

So the model picked up on that rule e.g., "being in the running wheel ROI for low cumulative time is a good predictor for fast running". That wasn't true in the larger data - and I struggled to think of a scenario where cumulative time / entries could be a useful indicator of any behavior so removed it.

@hdsliu
Copy link
Author

hdsliu commented Aug 6, 2024

Sounds reasonable. The behaviors I am annotating (rearing and running) probably won't benefit from these features anyway.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants