Marketing Science in Python: Code Notebooks

Executable notebooks and helper code for the book Marketing Science in Python: A Practitioner's Guide by Hajime Takeda.

📖 Read the book: https://marketing-science-in-python.com/

What this repository is

This repository holds the runnable code for the book: the chapter notebooks, local helper modules, and the small datasets needed to reproduce the examples. Use it to run the notebooks and follow along with the book.

What's included

• notebooks/ — chapter notebooks (.ipynb), grouped by part
• src/msbook/ — local helper modules used by the notebooks for paths and dataset loading
• data/ — small committed and synthetic datasets, dataset profiles, and a guide to obtaining the larger public datasets (data/raw/README.md)
• environment.yml and per-part requirements.txt — environment setup
• pyproject.toml — installs the msbook local helper modules (pip install -e .)

The book manuscript and its build system live elsewhere; this repository is code only.

Getting started

# 1. Clone
git clone https://github.com/takechanman1228/marketing-science-in-python.git
cd marketing-science-in-python

# 2. Create the environment and install the local helper modules
conda env create -f environment.yml
conda activate marketing-science
pip install -e .

# 3. Open a notebook
jupyter lab notebooks/part3-causal-inference/sec3.2_ab_test_figures.ipynb

Prefer the browser? Every notebook has an Open in Colab badge at the top. If you use pip instead of conda, install the per-part requirements.txt for the chapter you are running, then pip install -e ..

Notebooks by part

• Part 3 — Causal Inference: A/B test analysis, quasi-experiments, meta-learners (CATE), uplift modeling
• Part 4 — Customer Analytics: segmentation (traditional and embedding-based), customer lifetime value with PyMC-Marketing (plus a LightGBM companion pipeline)
• Part 5 — Commercial Analytics: price elasticity, assortment optimization, demand forecasting
• Part 6 — Media Optimization: marketing mix modeling, end to end

Data

Small synthetic and derived datasets are committed under data/generated/. Larger public datasets (Dunnhumby, Iowa liquor sales, Online Retail II, MineThatData) are not redistributed here; data/raw/README.md explains how to obtain each one, and the msbook.data loaders read or fetch them from data/raw/.

License

• Code (notebooks, Python files, the msbook helper modules): MIT — see LICENSE-CODE.
• Book content and text: © 2025–2026 Hajime Takeda, all rights reserved — see LICENSE.

Citation

If you reference this work, please cite it using CITATION.cff.