Thinking with Data

A short book about the work that happens before any analysis, namely turning a vague request into a question that data can answer. Shron proposes a scoping framework covering the context you are working in, the need behind the request, the vision of what an answer looks like, and the outcome once somebody acts on it. He then covers arguments and evidence, borrowing from informal logic to show that an analysis is a chain of claims which can be attacked at any link. There are chapters on recurring patterns of analytical work, such as building a dimensional model or comparing against a baseline, and on closing a project cleanly. Read it before writing the query rather than after presenting the results.