Combine multiple lines from different columns and hovertool using bokeh [OP on SO]

Community Support
1

[I originally posted this question on [SO](https://stackoverflow.com/q/46199911/671013)]

Problem setting Assume I have a pandas.DataFrame which contains several columns and indexed by DatetimeIndex:

np.random.seed(42)
df = pd.DataFrame(np.random.random(size=(10,3)),
                  columns=['foo', 'bar', 'goo'],
                  index=pd.date_range('20170101', periods=10)
                 )

The objective is to plot the three columns, having the index as the x-axis and enable a hovertoolvertically.

Here’s what I managed to have so far:

hovertool = bmo.HoverTool(
    tooltips=[("index", "@index{%F}"),
              ("val", "$y{0,0.0}")],
    formatters={'index' : 'datetime'})
p = bpl.figure(width=900, height=400, x_axis_type="datetime", tools=[hovertool])
for i, col in enumerate(ds.column_names[:-1]):
    p.line('index', col, source=ds, color=Set2[4][i], legend=bcp.value(col))
bpl.show(p)

This works nicely and as expected (well depending on ones expectations). Anyways, if I replace it with:

hovertool = bmo.HoverTool(
    tooltips=[("index", "@index{%F}"),
              ("val", "$y{0,0.000}")],
    formatters={'index' : 'datetime'},
    mode='vline' # Enabling the vertical behavior
)
p = bpl.figure(width=900, height=400, x_axis_type="datetime", tools=[hovertool])
for i, col in enumerate(ds.column_names[:-1]):
    p.line('index', col, source=ds, color=Set2[4][i], legend=bcp.value(col))
bpl.show(p)

Then, two things doesn’t work:

  1. The labels are not aligned vertically (and show values for different x-positions)
  2. The value rendered is the one where the mouse is located, and not reflecting the right value of the other 2 curves.
2

I was not expecting it to be so complicated. Or is it that simple? :slight_smile:

···

On Thursday, September 14, 2017 at 8:28:50 AM UTC+2, Dror Atariah wrote:

[I originally posted this question on [SO](https://stackoverflow.com/q/46199911/671013)]

Problem setting Assume I have a pandas.DataFrame which contains several columns and indexed by DatetimeIndex:

np.random.seed(42)
df = pd.DataFrame(np.random.random(size=(10,3)),
                  columns=['foo', 'bar', 'goo'],
                  index=pd.date_range('20170101', periods=10)
                 )

The objective is to plot the three columns, having the index as the x-axis and enable a hovertoolvertically.

Here’s what I managed to have so far:

hovertool = bmo.HoverTool(
    tooltips=[("index", "@index{%F}"),
              ("val", "$y{0,0.0}")],
    formatters={'index' : 'datetime'})
p = bpl.figure(width=900, height=400, x_axis_type="datetime", tools=[hovertool])
for i, col in enumerate(ds.column_names[:-1]):
    p.line('index', col, source=ds, color=Set2[4][i], legend=bcp.value(col))
bpl.show(p)

This works nicely and as expected (well depending on ones expectations). Anyways, if I replace it with:

hovertool = bmo.HoverTool(
    tooltips=[("index", "@index{%F}"),
              ("val", "$y{0,0.000}")],
    formatters={'index' : 'datetime'},
    mode='vline' # Enabling the vertical behavior
)
p = bpl.figure(width=900, height=400, x_axis_type="datetime", tools=[hovertool])
for i, col in enumerate(ds.column_names[:-1]):
    p.line('index', col, source=ds, color=Set2[4][i], legend=bcp.value(col))
bpl.show(p)

Then, two things doesn’t work:

  1. The labels are not aligned vertically (and show values for different x-positions)
  2. The value rendered is the one where the mouse is located, and not reflecting the right value of the other 2 curves.