How to rotate X-axis labels in bokeh figure?

Liam Hanninen picture Liam Hanninen · Feb 20, 2017 · Viewed 14.6k times · Source

I'm just starting to use Bokeh. Below I create some args I use for the rect figure.

x_length = var_results.index * 5.5

Multiplying the index by 5.5 gave me more room between labels.

names = var_results.Feature.tolist()
y_length = var_results.Variance
y_center = var_results.Variance/2

var_results is a Pandas dataframe that has a typical, sequential, non-repeating index. var_results also has a column Features that is strings of non-repeated, names, and finally it has a column Variance which is dtype float.

r = figure(x_range = names, 
           y_range = (-0.05,.3), 
           active_scroll = 'wheel_zoom', 
           x_axis_label = 'Features', 
           y_axis_label = 'Variance')



r.rect(x_length, 
       y_center, 
       width=1, 
       height=y_length, 
       color = "#ff1200")
output_notebook()
show(r)

I'm essentially making a bar chart with rectangles. Bokeh seems to be very customizable. But my graph looks rough around the edges, literally.

enter image description here

As you can see there is an ugly smudge just below the chart and above the x-axis title 'Features'. This is the label titles (technically the rectangle titles). How do I create space for and perhaps rotate to 45 degrees the labels so that they are readable and not just an overlapping mess?

Answer

bluenote10 picture bluenote10 · Apr 2, 2017

In order to rotate the labels e.g. by 90 degrees to the left, you can set major_label_orientation to π/2. This can be done either when creating the axis element (as a kwarg to the axis constructor if you are using low level plotting) or also after you have created a plot/figure, for instance by:

p.xaxis.major_label_orientation = math.pi/2

# or alternatively:
p.xaxis.major_label_orientation = "vertical"

See also this example in the documentation.