Sliding window changes by mibur1 · Pull Request #33 · neurodatascience/dFC

mibur1 · 2025-02-14T10:23:18Z

I implemented the changes as discussed in #31.

The changes are:

Removed extending the window which has two effects:
- Bugfix regarding the zero-padding for rectangular SW calculation
- Limits the tapered window to contain no samples outside the window width
Added a window_std parameter to Sliding Window, Discrete HMM, and Sliding Window Clustering
- Default is None, which results to the original std of 3 * W / 22
Disabled the convolution and instead directly use the Gaussian
- In my opinion this is unnecessary now, as the window size is a hard limit for the window
Vectorized Pearson correlation calculation for improved performance

Feel free to modify this to your liking. Also I did not write any tests/documentation, please let me know if I should do so :)

mtorabi59 · 2025-02-17T22:06:23Z

Thanks @mibur1 ! Overall it looks good! Thank you for making the changes. There a few comments below:

pydfc/dfc_methods/sliding_window.py

mtorabi59 · 2025-02-17T22:46:52Z

@mibur1 lmk if you have any Qs!
Also, tests are not necessary, but if you have some ideas for testing the sliding window, it'd be great to have them!

Co-authored-by: MOHAMMAD TORABI <72619172+mtorabi59@users.noreply.github.com>

mibur1 · 2025-02-18T13:34:23Z

@mtorabi59 Thanks for the quick review and comments, I hope I addressed all of them. I will think about tests for the sliding window and add some if I have an idea.

mtorabi59 · 2025-02-18T18:41:27Z

@mibur1 Thanks, everything looks good! There might only be some formatting changes by black or flake8.

mtorabi59 · 2025-02-18T18:48:04Z

@mibur1 If you're familiar with pre-commit, it would be great to run it locally and make the required formatting changes, ow I'll fix them!

mibur1 · 2025-02-20T19:47:38Z

Thanks! I'll try to do this as well as the unit tests next week. But please feel free to do it yourself if you would like to have it done earlier.

mtorabi59 · 2025-02-25T15:00:05Z

@mibur1 Thanks !

mtorabi59 · 2025-03-04T19:07:48Z

@mibur1 Since the remaining changes are very small and you have done most of the work, I'm just gonna make the final adjustments myself.
Thanks again for your contribution!

mibur1 · 2025-03-05T15:54:15Z

Sorry, I forgot about that. Thanks a lot for handling it!

mibur1 added 2 commits February 14, 2025 10:55

added std parameter to SW methods and limit the width to the window size

1af1a72

vectorized pearson correlation

fd8db39

mtorabi59 self-requested a review February 17, 2025 22:02