This book is Work in Progress. I appreciate your feedback to make the book better.

Chapter 3 Unstructured Data

Pages, words, and places

Most of the information in the world was never poured into a spreadsheet. It sits on web pages, in sentences, on maps — shapes that no statistical function will accept. The task in this chapter is conversion: turning something that is not a table into one, without losing the thing that made it worth having.

That conversion is never neutral. Deciding which part of a page is data, which words count as the same word, or which coordinates belong to which region are analytical choices made before the analysis begins, and they deserve the same scrutiny as any model.