Appendix A: Python Visualization Setup Guide

This appendix walks you through setting up a complete Python data visualization environment. Whether you are starting from scratch or adding visualization packages to an existing workflow, follow the steps below to get every library used in this book installed and configured.

Choosing a Python Distribution

You have two mainstream options for managing Python installations:

Anaconda / Miniconda (recommended for this book). Conda handles both Python packages and non-Python system dependencies (GDAL, PROJ, and GEOS for geospatial work; C libraries for HDF5). If your work involves geospatial visualization (Chapter 23) or scientific computing (Chapter 27), conda will save you hours of troubleshooting compiled dependencies.

• Download Miniconda (lightweight) from https://docs.conda.io/en/latest/miniconda.html
• Or download the full Anaconda distribution from https://www.anaconda.com/download

pip + venv (lighter-weight alternative). If you prefer a minimal footprint and do not need geospatial libraries, standard pip with virtual environments works well. Use python -m venv to create isolated environments.

Creating an Environment

With conda:

conda create -n dataviz python=3.11 -y
conda activate dataviz

With pip:

python -m venv dataviz-env
source dataviz-env/bin/activate   # macOS/Linux
dataviz-env\Scripts\activate      # Windows

Core Packages

Install these packages first. They are used in nearly every chapter.

# conda
conda install matplotlib seaborn pandas numpy jupyterlab -y

# pip
pip install matplotlib seaborn pandas numpy jupyterlab

Interactive Visualization Packages

These libraries produce browser-based, interactive charts.

# conda
conda install -c conda-forge plotly altair vega_datasets -y

# pip
pip install plotly altair vega_datasets

For Plotly charts to render inside JupyterLab, you may also need:

pip install jupyterlab-plotly

Dashboard and Reporting Packages

pip install streamlit dash fpdf2 python-pptx

Streamlit and Dash are pip-only for the latest versions. Conda-forge packages exist but may lag behind.

Specialized Visualization Packages

For geospatial work, conda handles binary dependencies far more reliably:

conda install -c conda-forge geopandas folium -y

For the remaining specialized packages:

pip install networkx pyvis datashader holoviews wordcloud scipy

One-Shot Install

If you want everything at once, the repository includes a requirements.txt file:

pip install -r requirements.txt

Or with conda, use the provided environment.yml:

conda env create -f environment.yml
conda activate dataviz

Jupyter Setup

JupyterLab is the recommended interface for working through this book.

jupyter lab

Useful Extensions

• jupyterlab-plotly -- Renders Plotly figures inline.
• jupyterlab-code-formatter -- Auto-formats code with Black or isort.
• jupyterlab-git -- Git integration inside JupyterLab.

Install them via:

pip install jupyterlab-plotly jupyterlab-code-formatter jupyterlab-git

Notebook Best Practices

1. Set matplotlib to inline mode at the top of each notebook: %matplotlib inline
2. For higher-resolution figures in notebooks, add: %config InlineBackend.figure_format = 'retina'
3. Use %load_ext autoreload and %autoreload 2 when developing helper modules.

IDE Recommendations

If you prefer an IDE over Jupyter for script-based work:

• VS Code with the Python and Jupyter extensions. Supports inline figure rendering, variable exploration, and integrated terminals. The best all-around choice for mixed notebook and script workflows.
• PyCharm Professional includes built-in Scientific Mode with inline matplotlib rendering and a DataFrame viewer. The Community edition lacks notebook support.
• Spyder ships with Anaconda and provides a MATLAB-like experience with a variable explorer and inline plot pane. Good for scientists who prefer a traditional IDE layout.

Verifying Your Installation

Run this script to confirm that all key packages import correctly:

import matplotlib; print(f"matplotlib  {matplotlib.__version__}")
import seaborn;    print(f"seaborn     {seaborn.__version__}")
import plotly;     print(f"plotly      {plotly.__version__}")
import altair;     print(f"altair      {altair.__version__}")
import pandas;     print(f"pandas      {pandas.__version__}")
import numpy;      print(f"numpy       {numpy.__version__}")

# Optional -- will raise ImportError if not installed
try:
    import geopandas; print(f"geopandas   {geopandas.__version__}")
except ImportError:
    print("geopandas   not installed (needed for Ch 23)")

try:
    import streamlit; print(f"streamlit   {streamlit.__version__}")
except ImportError:
    print("streamlit   not installed (needed for Ch 29)")

print("\nAll core packages loaded successfully.")

Troubleshooting Common Issues

matplotlib figures not displaying. Make sure you run %matplotlib inline in Jupyter or call plt.show() in scripts.

Plotly blank output in JupyterLab. Install jupyterlab-plotly and restart the lab server.

geopandas installation fails with pip. Use conda instead: conda install -c conda-forge geopandas. The C library dependencies (GDAL, PROJ, GEOS) are pre-built in conda.

Conflicting package versions. Always work inside a dedicated virtual environment or conda environment. Avoid installing visualization libraries into your system Python.

Permission errors on Windows. Run your terminal as Administrator or use the --user flag: pip install --user matplotlib.

Apple Silicon (M1/M2/M3) compatibility. Conda-forge packages are built for ARM64. If a package installs but crashes, ensure you are using a native ARM conda rather than an x86 Rosetta environment.