April 6, 2020

Brian D. Ripley, Spatial Statistics. On the one hand, this is from 1981, so all the detailed computational advice is laughably obsolete. (At one point, Ripley discusses strategies for not having to keep all of a 128 kb image in main memory at once.) There has also been a lot of advances in some aspects of the theory, notably point processes. On the other hand, Ripley’s basic advice — visualize; do less testing for “randomness” and more model-building; simulate your models, visualize the simulations, and test modeling assumptions with simulations and visualizations; smooth, and remember that “kriging” is just the Wiener filter — remains eminently sound. — I have been reading bits and pieces of this book, off and on, since around 2000, but I have a rule about not recommending something until I’ve finished it completely. Having finally now read it all, including the chapter on tomography (!), I can safely say: anyone seriously interested in spatial statistics probably ought to read this, but you can skip the tomography chapter as obsolete. — Books to Read While the Algae Grow in Your Fur, March 2020