Marketing Science in Python
A Practitioner's Guide to Causal Inference, Marketing Mix Modeling, Pricing, Forecasting, and Customer Analytics
By Hajime Takeda, co-founder of Kuwalyst
Sales rose during last month's campaign. Did the campaign cause the increase, or was it warmer weather, or a competitor running out of stock? Marketing Science in Python is a practical guide to making marketing decisions more reliable. It shows how to frame a question that analysis can answer, how to estimate what a campaign actually caused, and how to use models and experiments to decide where the media budget goes. Every hands-on chapter in parts 3 to 6 comes with a Python notebook that runs on public or synthetic data.
Its author, Hajime Takeda, is a marketing scientist and co-founder of Kuwalyst. The book sets out the principles the Media Planner is built on: causal evidence for decisions made under uncertainty, media mix models calibrated with experiments, and a stated source for every number. It shows how we think, and it is free to read online.
What is inside
- Introduction: what marketing science is, and how to understand marketing data
- Analysis for better decisions: framing the right question, designing KPIs for diagnosis, turning data into a story
- Causal inference for marketing: A/B tests, quasi-experiments, causal impact with time series, uplift modelling
- Customer analytics: segmentation and customer lifetime value
- Commercial analytics: pricing, assortment and sales forecasting
- Media investment and optimisation: building an MMM, incrementality experiments, and calibrating the model with their results
The full text is free on the book's site, with the notebooks on GitHub.
Get the PDF
Prefer to read offline? Leave your details and we will email you a personal link to download the English PDF.