[ENH] Group shading

[ryanhammonds]

This is a point in #193 and adds shading around group spectra. By default, the shading defaults to +/- 1 stdev. This is bypassed if a 1d array is passed to shade.

import numpy as np
from neurodsp.sim import sim_combined
from neurodsp.spectral import compute_spectrum
from fooof.plts import plot_spectra
from fooof.plts.error import plot_error_shade

components = {'sim_bursty_oscillation': {'freq': 10},
              'sim_powerlaw': {'exponent': -2}}

n_iters = 20

powers = np.zeros((n_iters, 50))

for idx in range(n_iters):
    sig = sim_combined(10, 1000, components)
    freqs, power = compute_spectrum(sig, 1000, f_range=(1, 50))
    powers[idx] = power
    
plot_spectra(freqs, powers, log_powers=True)    
plot_error_shade(freqs, powers, log_powers=True)

[TomDonoghue]

I updated this func with some of the changes we chatted about recently including moving, renaming (though I'm still not sure about the name), and updating approaches for calculate average and shade. @ryanhammonds - can you check the updates, and see if this looks good to you, and test it out that everything works? Thanks!

[ryanhammonds]

I left a few comments that I'm going to push updates for. If you agree with the comments/changes then this should be ready to merge. The changes were related to getting custom callables working:

def average_callable(powers):
    
    return np.mean(powers, axis=0)

def shade_callable(powers):
    
    return np.std(powers, axis=0)

plot_spectra_yshade(freqs, powers, shade=shade_callable,
                    average=average_callable, log_powers=True)

[ryanhammonds]

This will require an average callable to have an axis arg. A custom callable may not have this, but np functions will. We may just want to require an array of length equal to the first dimension of plt_powers to be returned?

If a numpy callable is expected, we should make a note.

[TomDonoghue]

Yeh, great points. I think what you've done makes a ton of sense to address this.

[TomDonoghue]

@ryanhammonds - thanks for the review here! You are totally correct, I totally dropped the ball on actually implementing callable support lol (oops - too much late night coding). I like your updates here - merging this in.