Regression
with R & Python

Regression with R and Python is a wonderful and inspiring guide to learning from data rather than merely processing it. It shows how the proliferation of data becomes valuable only when statistical reasoning transforms information into knowledge. Its great strength is the learning path: from description and visualization, through prediction, to causal reasoning and credible empirical conclusions. R and Python turn this path into an active experience in which readers reproduce, modify, and extend real analyses. The book thereby connects data analytics with understanding: computation serves interpretation, rather than replacing it. Different learning depths—from intuitive applications to advanced technical material—allow readers to construct their own route through the subject. A timely book that teaches not only how to analyse data, but how data become knowledge.

Johansson and Sun have written a regression textbook that does something genuinely useful: it insists, from Chapter 1 onward, that the right way to analyse data depends on whether the question is descriptive, predictive, or causal, and it carries that distinction through every subsequent chapter rather than treating it as a one-off framing device. The nonparametric regression material is a particular strength. The exposition moves cleanly from the regressogram's intuitive local-averaging logic through kernel and local polynomial regression to the bias-variance trade-off and the subtleties of inference under smoothing, with the more technical asymptotics cleanly separated into a companion appendix so the main text stays accessible. The connections to regression discontinuity design and to robust HAC and self-normalized inference are handled with real care, reflecting the authors' own research in these areas. I would be glad to recommend this book to students and applied researchers who want to understand not just how to run a regression, but what it can and cannot tell them.