I'm trying to plot a time series forecast, surrounded by a shaded uncertainty interval timeseries, and have tooltips work for both the line and the perimeter of the uncertainty.
import pandas as pd
from bqplot import *
daterange = pd.date_range(start='2020-01-01', freq='1D', periods=20)
df = pd.DataFrame(index=daterange)
df['fcst'] = np.sin(np.arange(0,20)*2*np.pi / 20)
tt_ex = Tooltip(fields=['x', 'y' ], labels=['', ''], formats=["%B %Y", ',.2f'])
x_sc = DateScale()
y_sc = LinearScale()
fcst_vals = np.arange(0,20)*2*np.pi / 20
x_ax_fcst = Axis(scale=x_sc)
y_ax_fcst = Axis(scale=y_sc, orientation='vertical', tick_format='.2f')
fcst_uncertainty = Lines(x=[daterange.append(daterange[::-1])],
y=[((df['fcst']+0.2).append((df['fcst'][::-1]-0.2)))],
fill_colors=['blue'],
fill='inside',
marker = 'cross',
stroke_width=1,
close_path=True,
scales={'x': x_sc, 'y': y_sc},
tooltip=tt_ex)
fcst_uncertainty.fill_opacities = [0.2]
fcst_line = Lines(x=[daterange], y=[df['fcst']],
scales={'x': x_sc, 'y': y_sc},
marker='circle', colors=['blue'],
tooltip=tt_ex)
example_fig = Figure(marks=[
fcst_line,
fcst_uncertainty
], axes=[x_ax_fcst, y_ax_fcst])
display(example_fig)
But the fill is blocking the tooltip for the main timeseries that is inside of the filled region. Is there any easy way around this? If I remove the fill, it works as desired. But I want the fill. I tried making another Lines object without a tooltip interaction and having that be the filled object, but that didn't work either. Thanks!
You'll kick yourself...reorder the marks in the Figure call to place the uncertainty first, and the line second. The order of the list functions similarly to zorder in matplotlib.
example_fig = Figure(marks=[
fcst_uncertainty,
fcst_line,
], axes=[x_ax_fcst, y_ax_fcst])
Nice example by the way.