Features

[kujaku11]

This PR will allow the ability to create features for FC's.

There is some book keeping as well as additional code.

A prototype json for defining weighting schemes based on features of the data is in aurora issue 61. To realize this we will need:

• An mt_metadata representation of a general feature (it has a name or id), and it has parameters. These are not defined for the abstract base class, but instead are defined individually for each feature, such as "coherence", "multiple_coherence", "mahalanobis_distance", etc.
• Separate jsons and classes for each feature we may want to implement (we will use "coherence", & "impedance", as first tests for these.
• A mt_metadata weighting_curve, that controls how weights are assigned for a given feature
• Specific features to use for weights should be allowed to vary depending on the output variable of the regression (ex, ey, hz), so each output variable will be associated with a list of features, as well as a weighting_strategy_parameters, which maps to a keyword. The simplest case would be multipilcation, i.e. simply multiply the weights from each feature together, but the keyword can map to more complicated methods in future, possibly also involving weighting_strategy_parameters.
• Most features that take FCs as input will be output in frequency bands. The feature bands are in general not necessarily equal to the transfer function estimation bands, and because of this, they need metadata to specify how the bands will be defined. Here are some ways to do that
  • use the same bands as for TF estimation, with a minimum number of FCs (so if the band has only 1 FC, we widen it by one more FC to each side for example)
  • Use a variation of SigMT's parzen window scheme
  • provide a custom set of frequency bands

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

Attention: Patch coverage is 81.95021% with 87 lines in your changes missing coverage. Please review.

Project coverage is 84.86%. Comparing base (ba2323e) to head (fdd02d6).
Report is 35 commits behind head on patches.

Files with missing lines	Patch %	Lines
...er_functions/processing/aurora/decimation_level.py	67.56%	36 Missing ⚠️
...ions/processing/fourier_coefficients/decimation.py	57.14%	24 Missing ⚠️
...t_metadata/transfer_functions/processing/window.py	78.04%	9 Missing ⚠️
...adata/transfer_functions/processing/aurora/band.py	65.00%	7 Missing ⚠️
...fer_functions/processing/aurora/frequency_bands.py	93.05%	5 Missing ⚠️
...er_functions/processing/fourier_coefficients/fc.py	71.42%	2 Missing ⚠️
...nctions/processing/short_time_fourier_transform.py	86.66%	2 Missing ⚠️
...transfer_functions/processing/aurora/processing.py	0.00%	1 Missing ⚠️
...unctions/processing/test_time_series_decimation.py	90.90%	1 Missing ⚠️

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@             Coverage Diff             @@
##           patches     #246      +/-   ##
===========================================
- Coverage    84.98%   84.86%   -0.12%     
===========================================
  Files          270      274       +4     
  Lines        20515    20535      +20     
===========================================
- Hits         17435    17428       -7     
- Misses        3080     3107      +27     

Flag	Coverage Δ
tests	84.86% <81.95%> (-0.12%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.