Plot Power Spectrum Models

Plotting power spectrum models, directly from specparam objects.

In order to the get a qualitative sense of if the model is fitting well, and what the results look like, it can be useful to visualize power spectrum model reconstructions.

This example dives deeper into plotting model reconstructions, using the plot() method from a SpectralModel object, and explores options for tuning these these visualizations.

# Import matplotlib to help manage plotting
import matplotlib.pyplot as plt

# Import the model object
from specparam import SpectralModel

# Import simulation functions to create some example data
from specparam.sim import sim_power_spectrum

# Simulate an example power spectrum
freqs, powers = sim_power_spectrum([3, 50], [1, 1],
                                   [[9, 0.25, 0.5], [22, 0.1, 1.5], [25, 0.2, 1.]])

Plotting from model objects

The model object has a plot() method that can be used to visualize data and models available in the SpectralModel object.

# Initialize a model object, and add some data to it
fm = SpectralModel(verbose=False)
fm.add_data(freqs, powers)

Once you have added data to a model object, you can visualize the data using plot().

# Plot the data available in the object
fm.plot()

When the model is available, the plot() call also displays the full model reconstruction, in red.

# Fit the model, and plot the reconstruction
fm.fit()
fm.plot()

Plotting Aperiodic Components

As you can see above, the plot() call by default also plots the aperiodic component, in a dashed blue line.

You can toggle whether to display the aperiodic component with the plot_aperiodic parameter.

# Control whether to plot the aperiodic component
fm.plot(plot_aperiodic=False)

Plotting Periodic Components

By default the peaks are only visualized as parts of the full model fit.

However, in some cases it can be useful to more explicitly visualize individual peaks, including where they are and if and how they overlap.

To do so, you can use the plot_peaks parameter, passing in a string specifier of which approach you wish to use to visualize the peaks.

There are four options for visualizing peaks:

• ‘shade’ : shade in peaks

• ‘dot’ : add a line through the peak, with a dot at the top

• ‘outline’ : outline each peak

• ‘line’ : add a vertical line through the whole plot at peak locations

# Plotting Periodic Components
fig, axes = plt.subplots(2, 2, figsize=[16, 12])
peak_plots = ['shade', 'dot', 'outline', 'line']
for ax, peak_plot in zip(axes.flatten(), peak_plots):
    fm.plot(plot_peaks=peak_plot, add_legend=False, ax=ax)

Note that you can also combine different peak visualizations together.

This can be done by joining all requested peak visualization approaches, with dashes (-).

For example, as plot_peaks=’dot-outline-shade’.

# Combine peak representations
fm.plot(plot_aperiodic=True, plot_peaks='line-shade-outline', plt_log=False)

Which peak visualization is best depends on how you want to look at peaks, and what you want to check.

For example, for investigating possible peak overlaps, the outline approach may be the most useful. Or, across broader frequency ranges, it may be easier to visualize where peaks were fit with the full-length vertical lines.

# Simulate a new power spectrum, over a broader frequency region
freqs, powers = sim_power_spectrum([1, 150], [0, 10, 1.5],
                                   [[4, 0.25, 1], [12, 0.2, 1.5], [25, 0.1, 2]])

# Initialize a new model object, in 'knee' mode
fm = SpectralModel(aperiodic_mode='knee', verbose=False)

# Fit the model and visualize the fit, highlighting peaks
fm.fit(freqs, powers)
fm.plot(plot_peaks='shade-line')

Other Plotting Options

There are also some other optional inputs to the plot() call, including:

• plt_log : Optional input for plotting the frequency axis in log10 spacing

• add_legend : Optional input to toggle whether to add a legend

• save_fig : Optional input for whether to save out the figure

  • you can control the filename and where to save to with file_name and file_path

• ax : Optional input to specify a matplotlib axes object to plot to

# Plot from specparam, using a custom axes with some optional inputs to tune the plot
_, ax = plt.subplots(figsize=[10, 10])
fm.plot(plt_log=True, add_legend=False, ax=ax)