def plot_spectra(freqs,
power_spectra,
log_freqs=False,
log_powers=False,
colors=None,
labels=None,
ax=None,
**plot_kwargs):
"""Plot one or multiple power spectra.
Parameters
----------
freqs : 1d or 2d array or list of 1d array
Frequency values, to be plotted on the x-axis.
power_spectra : 1d or 2d array or list of 1d array
Power values, to be plotted on the y-axis.
log_freqs : bool, optional, default: False
Whether to plot the frequency axis in log spacing.
log_powers : bool, optional, default: False
Whether to plot the power axis in log spacing.
colors : list of str, optional, default: None
Line colors of the spectra.
labels : list of str, optional, default: None
Legend labels for the spectra.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
**plot_kwargs
Keyword arguments to pass into the ``style_plot``.
"""
ax = check_ax(ax, plot_kwargs.pop('figsize', PLT_FIGSIZES['spectral']))
# Create the plot
plot_kwargs = check_plot_kwargs(plot_kwargs, {'linewidth': 2.0})
# Make inputs iterable if need to be passed multiple times to plot each spectrum
plt_powers = np.reshape(power_spectra, (1, -1)) if np.ndim(power_spectra) == 1 else \
power_spectra
plt_freqs = repeat(freqs) if isinstance(
freqs, np.ndarray) and freqs.ndim == 1 else freqs
# Set labels
labels = plot_kwargs.pop('label') if 'label' in plot_kwargs.keys(
) and labels is None else labels
labels = repeat(labels) if not isinstance(labels, list) else cycle(labels)
colors = repeat(colors) if not isinstance(colors, list) else cycle(colors)
# Plot
for freqs, powers, color, label in zip(plt_freqs, plt_powers, colors,
labels):
# Set plot data, logging if requested, and collect color, if absent
freqs = np.log10(freqs) if log_freqs else freqs
powers = np.log10(powers) if log_powers else powers
if color:
plot_kwargs['color'] = color
ax.plot(freqs, powers, label=label, **plot_kwargs)
style_spectrum_plot(ax, log_freqs, log_powers)
def plot_spectra_shading(freqs,
power_spectra,
shades,
shade_colors='r',
add_center=False,
ax=None,
**plot_kwargs):
"""Plot one or multiple power spectra with a shaded frequency region (or regions).
Parameters
----------
freqs : 1d or 2d array or list of 1d array
Frequency values, to be plotted on the x-axis.
power_spectra : 1d or 2d array or list of 1d array
Power values, to be plotted on the y-axis.
shades : list of [float, float] or list of list of [float, float]
Shaded region(s) to add to plot, defined as [lower_bound, upper_bound].
shade_colors : str or list of string
Color(s) to plot shades.
add_center : bool, optional, default: False
Whether to add a line at the center point of the shaded regions.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
**plot_kwargs
Keyword arguments to pass into :func:`~.plot_spectra`.
Notes
-----
Parameters for `plot_spectra` can also be passed into this function as keyword arguments.
This includes `log_freqs`, `log_powers` & `labels`. See `plot_spectra` for usage details.
"""
ax = check_ax(ax, plot_kwargs.pop('figsize', PLT_FIGSIZES['spectral']))
plot_spectra(freqs, power_spectra, ax=ax, **plot_kwargs)
add_shades(ax, shades, shade_colors, add_center,
plot_kwargs.get('log_freqs', False))
style_spectrum_plot(ax, plot_kwargs.get('log_freqs', False),
plot_kwargs.get('log_powers', False))
def plot_spectral_error(freqs,
error,
shade=None,
log_freqs=False,
ax=None,
**plot_kwargs):
"""Plot frequency by frequency error values.
Parameters
----------
freqs : 1d array
Frequency values, to be plotted on the x-axis.
error : 1d array
Calculated error values or mean error values across frequencies, to plot on the y-axis.
shade : 1d array, optional
Values to shade in around the plotted error.
This could be, for example, the standard deviation of the errors.
log_freqs : bool, optional, default: False
Whether to plot the frequency axis in log spacing.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
**plot_kwargs
Keyword arguments to pass into the ``style_plot``.
"""
ax = check_ax(ax, plot_kwargs.pop('figsize', PLT_FIGSIZES['spectral']))
plt_freqs = np.log10(freqs) if log_freqs else freqs
plot_spectra(plt_freqs, error, ax=ax, linewidth=3)
if np.any(shade):
ax.fill_between(plt_freqs, error - shade, error + shade, alpha=0.25)
ymin, ymax = ax.get_ylim()
if ymin < 0:
ax.set_ylim([0, ymax])
ax.set_xlim(plt_freqs.min(), plt_freqs.max())
style_spectrum_plot(ax, log_freqs, True)
ax.set_ylabel('Absolute Error')
def plot_fm(fm,
plot_peaks=None,
plot_aperiodic=True,
plt_log=False,
add_legend=True,
save_fig=False,
file_name=None,
file_path=None,
ax=None,
data_kwargs=None,
model_kwargs=None,
aperiodic_kwargs=None,
peak_kwargs=None,
**plot_kwargs):
"""Plot the power spectrum and model fit results from a FOOOF object.
Parameters
----------
fm : FOOOF
Object containing a power spectrum and (optionally) results from fitting.
plot_peaks : None or {'shade', 'dot', 'outline', 'line'}, optional
What kind of approach to take to plot peaks. If None, peaks are not specifically plotted.
Can also be a combination of approaches, separated by '-', for example: 'shade-line'.
plot_aperiodic : boolean, optional, default: True
Whether to plot the aperiodic component of the model fit.
plt_log : boolean, optional, default: False
Whether to plot the frequency values in log10 spacing.
add_legend : boolean, optional, default: False
Whether to add a legend describing the plot components.
save_fig : bool, optional, default: False
Whether to save out a copy of the plot.
file_name : str, optional
Name to give the saved out file.
file_path : str, optional
Path to directory to save to. If None, saves to current directory.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
data_kwargs, model_kwargs, aperiodic_kwargs, peak_kwargs : None or dict, optional
Keyword arguments to pass into the plot call for each plot element.
**plot_kwargs
Keyword arguments to pass into the ``style_plot``.
Notes
-----
Since FOOOF objects store power values in log spacing,
the y-axis (power) is plotted in log spacing by default.
"""
ax = check_ax(ax, PLT_FIGSIZES['spectral'])
# Log settings - note that power values in FOOOF objects are already logged
log_freqs = plt_log
log_powers = False
# Plot the data, if available
if fm.has_data:
data_defaults = {
'color': PLT_COLORS['data'],
'linewidth': 2.0,
'label': 'Original Spectrum' if add_legend else None
}
data_kwargs = check_plot_kwargs(data_kwargs, data_defaults)
plot_spectra(fm.freqs,
fm.power_spectrum,
log_freqs,
log_powers,
ax=ax,
**data_kwargs)
# Add the full model fit, and components (if requested)
if fm.has_model:
model_defaults = {
'color': PLT_COLORS['model'],
'linewidth': 3.0,
'alpha': 0.5,
'label': 'Full Model Fit' if add_legend else None
}
model_kwargs = check_plot_kwargs(model_kwargs, model_defaults)
plot_spectra(fm.freqs,
fm.fooofed_spectrum_,
log_freqs,
log_powers,
ax=ax,
**model_kwargs)
# Plot the aperiodic component of the model fit
if plot_aperiodic:
aperiodic_defaults = {
'color': PLT_COLORS['aperiodic'],
'linewidth': 3.0,
'alpha': 0.5,
'linestyle': 'dashed',
'label': 'Aperiodic Fit' if add_legend else None
}
aperiodic_kwargs = check_plot_kwargs(aperiodic_kwargs,
aperiodic_defaults)
plot_spectra(fm.freqs,
fm._ap_fit,
log_freqs,
log_powers,
ax=ax,
**aperiodic_kwargs)
# Plot the periodic components of the model fit
if plot_peaks:
_add_peaks(fm, plot_peaks, plt_log, ax, peak_kwargs)
# Apply default style to plot
style_spectrum_plot(ax, log_freqs, True)
def plot_spectra_yshade(freqs,
power_spectra,
shade='std',
average='mean',
scale=1,
log_freqs=False,
log_powers=False,
color=None,
label=None,
ax=None,
**plot_kwargs):
"""Plot standard deviation or error as a shaded region around the mean spectrum.
Parameters
----------
freqs : 1d array
Frequency values, to be plotted on the x-axis.
power_spectra : 1d or 2d array
Power values, to be plotted on the y-axis. ``shade`` must be provided if 1d.
shade : 'std', 'sem', 1d array or callable, optional, default: 'std'
Approach for shading above/below the mean spectrum.
average : 'mean', 'median' or callable, optional, default: 'mean'
Averaging approach for the average spectrum to plot. Only used if power_spectra is 2d.
scale : int, optional, default: 1
Factor to multiply the plotted shade by.
log_freqs : bool, optional, default: False
Whether to plot the frequency axis in log spacing.
log_powers : bool, optional, default: False
Whether to plot the power axis in log spacing.
color : str, optional, default: None
Line color of the spectrum.
label : str, optional, default: None
Legend label for the spectrum.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
**plot_kwargs
Keyword arguments to be passed to `plot_spectra` or to the plot call.
"""
if (isinstance(shade, str)
or isfunction(shade)) and power_spectra.ndim != 2:
raise ValueError('Power spectra must be 2d if shade is not given.')
ax = check_ax(ax, plot_kwargs.pop('figsize', PLT_FIGSIZES['spectral']))
# Set plot data & labels, logging if requested
plt_freqs = np.log10(freqs) if log_freqs else freqs
plt_powers = np.log10(power_spectra) if log_powers else power_spectra
# Organize mean spectrum to plot
avg_funcs = {'mean': np.mean, 'median': np.median}
if isinstance(average, str) and plt_powers.ndim == 2:
avg_powers = avg_funcs[average](plt_powers, axis=0)
elif isfunction(average) and plt_powers.ndim == 2:
avg_powers = average(plt_powers)
else:
avg_powers = plt_powers
# Plot average power spectrum
ax.plot(plt_freqs, avg_powers, linewidth=2.0, color=color, label=label)
# Organize shading to plot
shade_funcs = {'std': np.std, 'sem': sem}
if isinstance(shade, str):
shade_vals = scale * shade_funcs[shade](plt_powers, axis=0)
elif isfunction(shade):
shade_vals = scale * shade(plt_powers)
else:
shade_vals = scale * shade
upper_shade = avg_powers + shade_vals
lower_shade = avg_powers - shade_vals
# Plot +/- yshading around spectrum
alpha = plot_kwargs.pop('alpha', 0.25)
ax.fill_between(plt_freqs,
lower_shade,
upper_shade,
alpha=alpha,
color=color,
**plot_kwargs)
style_spectrum_plot(ax, log_freqs, log_powers)
def plot_annotated_peak_search(fm):
"""Plot a series of plots illustrating the peak search from a flattened spectrum.
Parameters
----------
fm : FOOOF
FOOOF object, with model fit, data and settings available.
"""
# Recalculate the initial aperiodic fit and flattened spectrum that
# is the same as the one that is used in the peak fitting procedure
flatspec = fm.power_spectrum - \
gen_aperiodic(fm.freqs, fm._robust_ap_fit(fm.freqs, fm.power_spectrum))
# Calculate ylims of the plot that are scaled to the range of the data
ylims = [
min(flatspec) - 0.1 * np.abs(min(flatspec)),
max(flatspec) + 0.1 * max(flatspec)
]
# Loop through the iterative search for each peak
for ind in range(fm.n_peaks_ + 1):
# This forces the creation of a new plotting axes per iteration
ax = check_ax(None, PLT_FIGSIZES['spectral'])
plot_spectra(fm.freqs,
flatspec,
ax=ax,
linewidth=2.5,
label='Flattened Spectrum',
color=PLT_COLORS['data'])
plot_spectra(fm.freqs,
[fm.peak_threshold * np.std(flatspec)] * len(fm.freqs),
ax=ax,
label='Relative Threshold',
color='orange',
linewidth=2.5,
linestyle='dashed')
plot_spectra(fm.freqs, [fm.min_peak_height] * len(fm.freqs),
ax=ax,
label='Absolute Threshold',
color='red',
linewidth=2.5,
linestyle='dashed')
maxi = np.argmax(flatspec)
ax.plot(fm.freqs[maxi],
flatspec[maxi],
'.',
color=PLT_COLORS['periodic'],
alpha=0.75,
markersize=30)
ax.set_ylim(ylims)
ax.set_title('Iteration #' + str(ind + 1), fontsize=16)
if ind < fm.n_peaks_:
gauss = gaussian_function(fm.freqs, *fm.gaussian_params_[ind, :])
plot_spectra(fm.freqs,
gauss,
ax=ax,
label='Gaussian Fit',
color=PLT_COLORS['periodic'],
linestyle=':',
linewidth=3.0)
flatspec = flatspec - gauss
style_spectrum_plot(ax, False, True)
def plot_annotated_model(fm,
plt_log=False,
annotate_peaks=True,
annotate_aperiodic=True,
ax=None):
"""Plot a an annotated power spectrum and model, from a FOOOF object.
Parameters
----------
fm : FOOOF
FOOOF object, with model fit, data and settings available.
plt_log : boolean, optional, default: False
Whether to plot the frequency values in log10 spacing.
annotate_peaks : boolean, optional, default: True
Whether to annotate the periodic components of the model fit.
annotate_aperiodic : boolean, optional, default: True
Whether to annotate the aperiodic components of the model fit.
ax : matplotlib.Axes, optional
Figure axes upon which to plot.
Raises
------
NoModelError
If there are no model results available to plot.
"""
# Check that model is available
if not fm.has_model:
raise NoModelError("No model is available to plot, can not proceed.")
# Settings
fontsize = 15
lw1 = 4.0
lw2 = 3.0
ms1 = 12
# Create the baseline figure
ax = check_ax(ax, PLT_FIGSIZES['spectral'])
fm.plot(plot_peaks='dot-shade-width',
plt_log=plt_log,
ax=ax,
data_kwargs={
'lw': lw1,
'alpha': 0.6
},
aperiodic_kwargs={
'lw': lw1,
'zorder': 10
},
model_kwargs={
'lw': lw1,
'alpha': 0.5
},
peak_kwargs={
'dot': {
'color': PLT_COLORS['periodic'],
'ms': ms1,
'lw': lw2
},
'shade': {
'color': PLT_COLORS['periodic']
},
'width': {
'color': PLT_COLORS['periodic'],
'alpha': 0.75,
'lw': lw2
}
})
# Get freqs for plotting, and convert to log if needed
freqs = fm.freqs if not plt_log else np.log10(fm.freqs)
## Buffers: for spacing things out on the plot (scaled by plot values)
x_buff1 = max(freqs) * 0.1
x_buff2 = max(freqs) * 0.25
y_buff1 = 0.15 * np.ptp(ax.get_ylim())
shrink = 0.1
# There is a bug in annotations for some perpendicular lines, so add small offset
# See: https://github.com/matplotlib/matplotlib/issues/12820. Fixed in 3.2.1.
bug_buff = 0.000001
if annotate_peaks and fm.n_peaks_:
# Extract largest peak, to annotate, grabbing gaussian params
gauss = get_band_peak_fm(fm,
fm.freq_range,
attribute='gaussian_params')
peak_ctr, peak_hgt, peak_wid = gauss
bw_freqs = [
peak_ctr - 0.5 * compute_fwhm(peak_wid),
peak_ctr + 0.5 * compute_fwhm(peak_wid)
]
if plt_log:
peak_ctr = np.log10(peak_ctr)
bw_freqs = np.log10(bw_freqs)
peak_top = fm.power_spectrum[nearest_ind(freqs, peak_ctr)]
# Annotate Peak CF
ax.annotate('Center Frequency',
xy=(peak_ctr, peak_top),
xytext=(peak_ctr, peak_top + np.abs(0.6 * peak_hgt)),
verticalalignment='center',
horizontalalignment='center',
arrowprops=dict(facecolor=PLT_COLORS['periodic'],
shrink=shrink),
color=PLT_COLORS['periodic'],
fontsize=fontsize)
# Annotate Peak PW
ax.annotate('Power',
xy=(peak_ctr, peak_top - 0.3 * peak_hgt),
xytext=(peak_ctr + x_buff1, peak_top - 0.3 * peak_hgt),
verticalalignment='center',
arrowprops=dict(facecolor=PLT_COLORS['periodic'],
shrink=shrink),
color=PLT_COLORS['periodic'],
fontsize=fontsize)
# Annotate Peak BW
bw_buff = (peak_ctr - bw_freqs[0]) / 2
ax.annotate('Bandwidth',
xy=(peak_ctr - bw_buff + bug_buff,
peak_top - (0.5 * peak_hgt)),
xytext=(peak_ctr - bw_buff, peak_top - (1.5 * peak_hgt)),
verticalalignment='center',
horizontalalignment='right',
arrowprops=dict(facecolor=PLT_COLORS['periodic'],
shrink=shrink),
color=PLT_COLORS['periodic'],
fontsize=fontsize,
zorder=20)
if annotate_aperiodic:
# Annotate Aperiodic Offset
# Add a line to indicate offset, without adjusting plot limits below it
ax.set_autoscaley_on(False)
ax.plot([freqs[0], freqs[0]],
[ax.get_ylim()[0], fm.fooofed_spectrum_[0]],
color=PLT_COLORS['aperiodic'],
linewidth=lw2,
alpha=0.5)
ax.annotate('Offset',
xy=(freqs[0] + bug_buff, fm.power_spectrum[0] - y_buff1),
xytext=(freqs[0] - x_buff1,
fm.power_spectrum[0] - y_buff1),
verticalalignment='center',
horizontalalignment='center',
arrowprops=dict(facecolor=PLT_COLORS['aperiodic'],
shrink=shrink),
color=PLT_COLORS['aperiodic'],
fontsize=fontsize)
# Annotate Aperiodic Knee
if fm.aperiodic_mode == 'knee':
# Find the knee frequency point to annotate
knee_freq = compute_knee_frequency(
fm.get_params('aperiodic', 'knee'),
fm.get_params('aperiodic', 'exponent'))
knee_freq = np.log10(knee_freq) if plt_log else knee_freq
knee_pow = fm.power_spectrum[nearest_ind(freqs, knee_freq)]
# Add a dot to the plot indicating the knee frequency
ax.plot(knee_freq,
knee_pow,
'o',
color=PLT_COLORS['aperiodic'],
ms=ms1 * 1.5,
alpha=0.7)
ax.annotate('Knee',
xy=(knee_freq, knee_pow),
xytext=(knee_freq - x_buff2, knee_pow - y_buff1),
verticalalignment='center',
arrowprops=dict(facecolor=PLT_COLORS['aperiodic'],
shrink=shrink),
color=PLT_COLORS['aperiodic'],
fontsize=fontsize)
# Annotate Aperiodic Exponent
mid_ind = int(len(freqs) / 2)
ax.annotate('Exponent',
xy=(freqs[mid_ind], fm.power_spectrum[mid_ind]),
xytext=(freqs[mid_ind] - x_buff2,
fm.power_spectrum[mid_ind] - y_buff1),
verticalalignment='center',
arrowprops=dict(facecolor=PLT_COLORS['aperiodic'],
shrink=shrink),
color=PLT_COLORS['aperiodic'],
fontsize=fontsize)
# Apply style to plot & tune grid styling
style_spectrum_plot(ax, plt_log, True)
ax.grid(True, alpha=0.5)
# Add labels to plot in the legend
da_patch = mpatches.Patch(color=PLT_COLORS['data'], label='Original Data')
ap_patch = mpatches.Patch(color=PLT_COLORS['aperiodic'],
label='Aperiodic Parameters')
pe_patch = mpatches.Patch(color=PLT_COLORS['periodic'],
label='Peak Parameters')
mo_patch = mpatches.Patch(color=PLT_COLORS['model'], label='Full Model')
handles = [
da_patch, ap_patch if annotate_aperiodic else None,
pe_patch if annotate_peaks else None, mo_patch
]
handles = [el for el in handles if el is not None]
ax.legend(handles=handles, handlelength=1, fontsize='x-large')