How can I do bins and Histograms in Bokeh?

Hi,

I’m back to using Bokeh from another plotting platform and I’m completely lost. I have a very simple example that should bin plotting bins as histograms, but I can’t seem to be able to get the bokeh API to work.

import pandas as pd
import numpy as np

def construct_labels(start, end):
    labels = []
    for index, x in enumerate(start):
        y = end[index]
        labels.append('({}, {}]'.format(x, y))
    return labels


values = {'Length': np.random.uniform(0, 4, 10)}
df = pd.DataFrame(values, columns=['Length'])

bin_step_size = 0.5

# List of bin points.
p_bins = np.arange(0, (df['Length'].max() + bin_step_size), bin_step_size)

# Reduce the tail to create the left side bounds.
p_left_limits = p_bins[:-1].copy()

# Cut the head to create the right side bounds.
p_right_limits = np.delete(p_bins, 0)

# Create the bins.
p_range_bins = pd.IntervalIndex.from_arrays(p_left_limits, p_right_limits)

# Create labels.
p_range_labels = construct_labels(p_left_limits, p_right_limits)

p_ranges_binned = pd.cut(
    df['Length'],
    p_range_bins,
    labels=p_range_labels,
    precision=0,
    include_lowest=True)

counts = p_ranges_binned.value_counts(sort=False)

… and from here I’m pretty much lost. I can plot the above in matplotlib now, and it looks like this. But I just can’t figure out what was the Bokeh way

Figure_11000×500 6.8 KB

Have you seen the examples at Handling categorical data — Bokeh 2.4.2 Documentation?
Have you tried modifying one of them to your use case?

I started with this one, and I got a bit lost: bar_colormapped.py — Bokeh 2.2.0 Documentation

Let me try this one. Thanks!

Got it, thank you!

done1239×891 27.9 KB

output_file("bars.html")
out = p_ranges_binned
counts = out.value_counts(sort=False)

bin_labels = p_range_labels

p = figure(x_range=bin_labels, plot_height=250, title="Range Counts",
           toolbar_location=None, tools="")

p.vbar(x=bin_labels, top=counts, width=0.9)

p.xgrid.grid_line_color = None
p.y_range.start = 0

show(p)

If you actually want histogram (which has a continuous axis) is there a reason you are implementing it with bar chart (which has a categorical axis)? That is bound to make things more complicated than necessary.

This is how I would normally draw a histogram:

hist, edges = np.histogram(data, density=True)

p = figure()
p.quad(top=hist, left=edges[:-1], right=edges[1:], line_color="white")

Edit: updated to use quad

Let me try this. I have already build the bins, but let me try if I can change things a bit to use your example.

Just to be clear, the fundamental difference is not how the bin are built (e.g. np.histrogram or somethign else, it doesn’t matter). But whether you pass a list of string (categorical) labels as the range, instead of just using the bar edge coordinates.

I’m getting an error:

Apologies, should be quad not vbar (answer updated)

Cool, seems to be working. Thank you.

Ideally, I will need to have percentages on the left (y-scale), is there a built in way, or should I rework that myself? I would like to show, like %5 instead of the count. Indicating that %5 of the values are in bin (0.5, 1.5]

There are lots of options for formatting tick labels:

I’d suggest giving things a try and asking a (new, separate) question if you get stuck

Sure, thank you. Could you please help me to find the documentation for the quad() method, is it this one?

Yes, also described in the users guide