Skillversitas

Environment Setup

Before plotting anything, you need Bokeh installed in a clean Python environment. Bokeh 3.x supports Python 3.10 or later and has few required dependencies, so installation is quick. The recommended arrangement is a virtual environment dedicated to the project, which keeps Bokeh versions from clashing with other visualization libraries.

Installing Bokeh

Install with pip or conda:

pip install bokeh

or, if you use Anaconda:

conda install bokeh

Verify the installation by printing the version from Python:

import bokeh
print(bokeh.__version__)  # e.g. 3.4.0

Anything at or above 3.0 is fine for this tutorial; older 2.x code uses slightly different signatures for some properties, so avoid mixing versions.

Creating a Virtual Environment

python3 -m venv bokeh-env
source bokeh-env/bin/activate
pip install bokeh pandas numpy jupyter

Pandas and NumPy are not hard dependencies for standalone plotting, but nearly every real workflow passes DataFrames or arrays to a ColumnDataSource, so install them together.

Checking the Samples

Bokeh ships a gallery of example apps. Running one confirms both the library and the Bokeh Server work:

bokeh sampledata
bokeh serve --show examples/app/sliders.py

The first command downloads optional sample datasets such as periodic_table and glucose. The second starts a local server and opens the sliders dashboard in your browser at http://localhost:5006/sliders.

Development Workflow

A typical setup uses three artifacts:

  • a virtual environment with bokeh, pandas, and numpy
  • a scratch notebook or script for quick figure experiments with show(p)
  • an application module such as app.py whose root is the curdoc() document, served with bokeh serve app.py

For standalone HTML output, calling show() on a figure writes a file next to the script and opens the default browser, so no web framework is required to preview results.

Common Installation Issues

  • Old Bokeh at import time — the active environment may not be the one you installed into. Run pip show bokeh and check the Location, or use python -m pip install bokeh inside the environment.
  • Port already in use — Bokeh Server defaults to port 5006. Start it on another port with bokeh serve app.py --port 5007.
  • Blank plots in the browser — verify BokehJS actually loaded; when you use bokeh.embed manually you must either reference the CDN or install local static resources with bokeh.resources.

Summary

Install Bokeh in a dedicated virtual environment with pandas and numpy, confirm the version with bokeh.__version__, and exercise the bundled sample apps to validate both standalone rendering and the Bokeh Server before writing your first figure.