bokeh-visualization-guide

Bokeh Visualization Guide

Overview

Bokeh is a Python library for creating interactive visualizations for modern web browsers, with over 20K stars on GitHub. Developed and maintained by NumFocus, Bokeh generates standalone HTML documents or serves live interactive applications. Its architecture renders graphics in the browser using BokehJS, meaning the resulting visualizations are portable and can be shared as static HTML files without requiring Python on the viewer's end.

For researchers, Bokeh offers a unique advantage: its server-backed interactive applications allow real-time data exploration during analysis. Unlike static plotting libraries, Bokeh lets researchers build tools where they can brush-select data points, link multiple views of the same dataset, and stream live data from instruments or simulations. This makes it invaluable for exploratory data analysis in laboratory and computational research settings.

Bokeh provides multiple levels of API access. The high-level bokeh.plotting interface is comparable in convenience to matplotlib, while the low-level bokeh.models interface gives fine-grained control over every visual element. The library also integrates with HoloViews and Panel for building complex dashboards with minimal code.

Getting Started with Bokeh

Installation and Basic Setup

# Install bokeh
# pip install bokeh

from bokeh.plotting import figure, show, output_file, output_notebook
from bokeh.models import ColumnDataSource, HoverTool
import numpy as np
import pandas as pd

# For Jupyter notebooks
output_notebook()

# For standalone HTML files
output_file("research_figure.html")

Basic Scatter Plot for Experimental Data

from bokeh.plotting import figure, show
from bokeh.models import ColumnDataSource, HoverTool

# Prepare data
data = pd.DataFrame({
    'sample_id': [f'S{i:03d}' for i in range(100)],
    'measurement_a': np.random.normal(5, 1.5, 100),
    'measurement_b': np.random.normal(10, 2, 100),
    'group': np.random.choice(['Control', 'Treatment A', 'Treatment B'], 100),
    'pvalue': np.random.uniform(0.001, 0.1, 100)
})

source = ColumnDataSource(data)

# Color mapping by group
color_map = {'Control': '#6B7280', 'Treatment A': '#3B82F6', 'Treatment B': '#EF4444'}
data['color'] = data['group'].map(color_map)

p = figure(
    title='Measurement A vs B by Treatment Group',
    x_axis_label='Measurement A (units)',
    y_axis_label='Measurement B (units)',
    width=700, height=500,
    tools='pan,wheel_zoom,box_zoom,reset,save'
)

for group, color in color_map.items():
    subset = data[data['group'] == group]
    p.circle(
        x='measurement_a', y='measurement_b',
        source=ColumnDataSource(subset),
        color=color, size=8, alpha=0.7,
        legend_label=group
    )

# Add hover tooltip
hover = HoverTool(tooltips=[
    ('Sample', '@sample_id'),
    ('Group', '@group'),
    ('Measure A', '@measurement_a{0.3f}'),
    ('Measure B', '@measurement_b{0.3f}'),
    ('p-value', '@pvalue{0.4f}')
])
p.add_tools(hover)
p.legend.location = 'top_left'
p.legend.click_policy = 'hide'

show(p)