When plotting a Sequence with lots of onsets, not all lines are drawn about thebeat HOT 8 CLOSED

The gaps should be filled up

from thebeat.

Was fixed in 9f0b9e4 for thebeat.helpers.plot_single_sequence, but not yet for thebeat.visualization.plot_multiple_sequences. Will take a look later because that's slightly different. The y-axis there is provided as a categorical variable, which matplotlib.pyplot.vlines does not support.

from thebeat.

Also, after 9f0b9e4 the lines have different widths. It's difficult to see but compare e.g. these two:

from thebeat.

I think this is the way forward:

import thebeat
import matplotlib.pyplot as plt
import numpy as np
import matplotlib.patches as mpatches

# Create sequence
s = thebeat.Sequence([100, 500, 1000] * 50, first_onset=20, name="test")

# Make empty fig and ax
fig, ax = plt.subplots()

# Set title
ax.set_title("Sequence")

# Get final IOI
final_ioi = s.iois[-1] if s.end_with_interval is True else 0

# Set x and y limits
ax.set_ylim(0, 1)
ax.set_xlim(0, s.onsets[-1] + final_ioi)

# Plot onsets as rectangles
for onset in s.onsets:
    rect = mpatches.Rectangle(xy=(onset, 0), height=1, width=np.min(s.iois) / 10, color="blue", fill=True)
    ax.add_patch(rect)

# Remove y tick labels
ax.set_yticklabels([])
ax.set_yticks([])

# Show plot
plt.show()

I am also closing #6, because that is related.

from thebeat.

Almost done in 9e23c3d . Solution is a bit hacky for now, that has to be improved in the future.

from thebeat.

I'm having trouble reproducing this issue. My observations:

• It does not seem to depend on the type of matplotlib version used
• When saving the figure in how originally these sequences were created, there didn't seem to be an issue, even at low dpi (e.g. 100)
• My guess is that it's a problem with the inline backend of matplotlib, as the first time this issue came to light was in a Jupyter notebook

So:

• We will revert commits d131cc8 and 9f0b9e4
• We will not merge the branch better_sequence_plots', but leave it be for a while. Let's see if the problem resurfaces at some point, but I think it was a pretty specific issue.
• We will take one idea from the new branch, namely another way of renaming the y-axis labels if multiple the same exist (completely unrelated to the pres. issue, here as a reminder). This will be done in a separate commit.

from thebeat.

9f0b9e4 reverted in cb76fb9

from thebeat.

Closing this issue.

from thebeat.