Search code examples
pythonplotlinkerbokehpandas-bokeh

Is it possible to link axes from different plots in pandas-bokeh package?

In regular Bokeh there is a nice feature of linking axes from different subplots in a grid which e.g. is very useful for zooming on multiple graphs simultaneously. But this doesn't seem to work in the pandas-bokeh package - any good solutions?

Solution

  • Yes it is.

    It is not streight forward, but you can make use of js_link of the x_range and this is not too complicated. The same of course for the y_range if wanted.

    Minimal Example

    import pandas as pd
import pandas_bokeh as pdb
from bokeh.plotting import output_notebook
output_notebook()

# data
data = list(range(5))
df = pd.DataFrame({'x':data, 'y':data[::-1]})

# plots
p1 = pdb.plot(df['x'], show_figure=False)
p2 = pdb.plot(df['y'], show_figure=False)

# linking p1 ranges to p2
p1.x_range.js_link('start', p2.x_range, 'start')
p1.x_range.js_link('end', p2.x_range, 'end')
p1.y_range.js_link('start', p2.y_range, 'start')
p1.y_range.js_link('end', p2.y_range, 'end')

# linking p2 ranges to p1 vice versa
p2.x_range.js_link('start', p1.x_range, 'start')
p2.x_range.js_link('end', p1.x_range, 'end')
p2.y_range.js_link('start', p1.y_range, 'start')
p2.y_range.js_link('end', p1.y_range, 'end')

pdb.plot_grid([[p1],[p2]])

    Comment

    In an ideal world

    p1 = pdb.plot(df['x'], show_figure=False)
p2 = pdb.plot(df['y'], x_arnge=p1.x_range, y_range=p1.y_range, show_figure=False)
pdb.plot_grid([[p1],[p2]])

    should work. This is not, because in pandas-bokeh.plot() not all keyword arguments of the figure object are evaluated. In my opinion this is a misconfiguration and I opend a ticket on GitHub to address this issue.