ENH: Simplify code of classification by oesteban · Pull Request #28 · ME-ICA/aroma

oesteban · 2020-11-11T16:49:12Z

Also, change the function name and signature to "predict(X)" to make it more similar to scikit-learn.
Also, remove runICA and the other function for registration.

Closes # .

Changes proposed in this pull request:

- Also, change the function name and signature to "predict(X)" to make it more similar to scikit-learn. - Also, remove runICA and the other function for registration.

eurunuela

The changes look good to me but should be complemented by an update in the CLI to make sure we have the ICA components. Same with aroma.py.

oesteban · 2020-11-12T08:59:28Z

Don't you prefer to go step by step with targeted PRs? We can deal with the CLI once we have a functional prototype.

oesteban · 2020-11-12T08:59:55Z

Once again, the CLI is literally the last thing I would care for :D

tsalo · 2020-11-12T16:53:35Z

-    features_df.to_csv(
-        op.join(out_dir, "classification_overview.txt"), sep="\t", index_label="IC"
-    )


Since this file is no longer written out here, it would be good to write it out in the workflow. We also need a corresponding change in the workflow function, I think.

handwerkerd · 2021-11-22T02:40:12Z

The core of the component classification code in tedana is that each classification step is its own function. I don't think it's realistic to completely harmonize the two sets of classification codes now, but if you set it up so that each classification decision is modularized, it should be realistic to use the same system for both in the near future.

eurunuela · 2021-11-23T13:39:33Z

-def write_metrics(features_df, out_dir, metric_metadata=None):
-    """Write out feature/classification information and metadata.
-
-    Parameters
-    ----------
-    features_df : (C x 5) :obj:`pandas.DataFrame`
-        DataFrame with metric values and classifications.
-        Must have the following columns: "edge_fract", "csf_fract", "max_RP_corr", "HFC", and
-        "classification".
-    out_dir : :obj:`str`
-        Output directory.
-    metric_metadata : :obj:`dict` or None, optional
-        Metric metadata in a dictionary.
-
-    Returns
-    -------
-    motion_ICs : array_like
-        Array containing the indices of the components identified as motion components.
-
-    Output
-    ------
-    AROMAnoiseICs.csv : A text file containing the indices of the
-                        components identified as motion components
-    desc-AROMA_metrics.tsv
-    desc-AROMA_metrics.json
-    """
-    # Put the indices of motion-classified ICs in a text file (starting with 1)
-    motion_ICs = features_df["classification"][features_df["classification"] == "rejected"].index
-    motion_ICs = motion_ICs.values
-
-    with open(op.join(out_dir, "AROMAnoiseICs.csv"), "w") as fo:
-        out_str = ",".join(motion_ICs.astype(str))
-        fo.write(out_str)
-
-    # Create a summary overview of the classification
-    out_file = op.join(out_dir, "desc-AROMA_metrics.tsv")
-    features_df.to_csv(out_file, sep="\t", index_label="IC")
-
-    if isinstance(metric_metadata, dict):
-        with open(op.join(out_dir, "desc-AROMA_metrics.json"), "w") as fo:
-            json.dump(metric_metadata, fo, sort_keys=True, indent=4)
-
-    return motion_ICs


Why do we want to remove this @tsalo @oesteban? I do not remember why it was removed.

This was very long ago, but it seems to me that the plan would be to write the outputs somewhere else, when we have more clarity on what we want to exactly write out.

Ok, I will create io.py and put it in there.

eurunuela · 2021-11-23T18:25:41Z

Ok guys, I've made the following changes:

The function to save the metrics write_metrics has been moved to io.py.
The classification is done in classification.py with a function for each criteria as @handwerkerd mentioned.

What do you think @tsalo @CesarCaballeroGaudes @oesteban?

Edit: no idea why the style check fails.

tsalo

My only problem is that now we're not tracking why each "bad" component is classified as such. I understand if we dropped "rationale" in favor of a list of tags, as discussed for tedana, but it looks like this information is just completely dropped.

tsalo · 2021-12-06T18:12:19Z

+HYPERPLANE = np.array([-19.9751070082159, 9.95127547670627, 24.8333160239175])
+
+
+def hfc_criteria(x, thr_hfc=THR_HFC):


Suggested change

def hfc_criteria(x, thr_hfc=THR_HFC):

def hfc_criterion(x, thr_hfc=THR_HFC):

Since it's just one criterion.

tsalo · 2021-12-06T18:12:53Z

+    :obj:`pandas.DataFrame`
+        Features table with additional column "classification".


Suggested change

:obj:`pandas.DataFrame`

Features table with additional column "classification".

:obj:`numpy.ndarray`

Classification (``True`` if the component is a CSF one).

Co-authored-by: Taylor Salo <tsalo006@fiu.edu>

tsalo reviewed Nov 11, 2020

View reviewed changes

Comment thread aroma/utils.py Outdated

ENH: Simplify code of classification

e38c5c8

- Also, change the function name and signature to "predict(X)" to make it more similar to scikit-learn. - Also, remove runICA and the other function for registration.

oesteban force-pushed the patch-1 branch from b051241 to e38c5c8 Compare November 11, 2020 17:39

eurunuela reviewed Nov 12, 2020

View reviewed changes

tsalo reviewed Nov 12, 2020

View reviewed changes

eurunuela mentioned this pull request Nov 19, 2021

Brainhack Donostia 2021 #54

Closed

eurunuela added 2 commits November 23, 2021 08:27

Merge remote-tracking branch 'upstream/main' into pr/28

1282268

Moved prediction to classification file.

9edc544

eurunuela reviewed Nov 23, 2021

View reviewed changes

eurunuela added 3 commits November 23, 2021 19:04

Updates to classification.py and io.py

1307fbf

Updated call to denoising function

c15a9bd

Removed breakpoint

1fce555

tsalo reviewed Dec 6, 2021

View reviewed changes

Update aroma/aroma.py

4b3141d

Co-authored-by: Taylor Salo <tsalo006@fiu.edu>

		HYPERPLANE = np.array([-19.9751070082159, 9.95127547670627, 24.8333160239175])


		def hfc_criteria(x, thr_hfc=THR_HFC):

	def hfc_criteria(x, thr_hfc=THR_HFC):
	def hfc_criterion(x, thr_hfc=THR_HFC):

		:obj:`pandas.DataFrame`
		Features table with additional column "classification".

Conversation

oesteban commented Nov 11, 2020

Uh oh!

Uh oh!

eurunuela left a comment

Choose a reason for hiding this comment

Uh oh!

oesteban commented Nov 12, 2020

Uh oh!

oesteban commented Nov 12, 2020

Uh oh!

tsalo Nov 12, 2020

Choose a reason for hiding this comment

Uh oh!

handwerkerd commented Nov 22, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

eurunuela Nov 23, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

oesteban Nov 23, 2021

Choose a reason for hiding this comment

Uh oh!

eurunuela Nov 23, 2021

Choose a reason for hiding this comment

Uh oh!

eurunuela commented Nov 23, 2021 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tsalo left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

tsalo Dec 6, 2021

Choose a reason for hiding this comment

Uh oh!

tsalo Dec 6, 2021

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

handwerkerd commented Nov 22, 2021 •

edited

Loading

eurunuela Nov 23, 2021 •

edited

Loading

eurunuela commented Nov 23, 2021 •

edited

Loading