Visual Statistics

Statistics with visuals instead of formulas

Introduction

Visual Statistics

Statistics you can see.

This book teaches statistics through pictures instead of formulas. It goes from data and distributions through tests, regression and machine learning to deep learning and time series. Every idea gets a figure made from real or simulated data, and some of the figures you can play with.

Contents

What data is, how to summarise it, what a model is, and how a graph can tell the truth or mislead.

The shapes data comes in: measurements, counts, extremes and several variables at once, and how to check which shape fits.

How sure can you be? Tests, intervals and the bootstrap, each drawn as a picture of chance.

Lines and curves through data: one predictor or many, yes or no outcomes, counts and grouped data.

How the data was collected decides what it can tell you: experiments, surveys, missing values and cause and effect.

Predicting from examples: classifiers, testing a model honestly, trees and forests, and explaining what a model does.

Finding structure when nobody gives you the answers: components, clusters and the odd ones out.

Neural networks, from one neuron to image classifiers, language models, diffusion and learning by trial and error.

Data in time order: trends, seasons and forecasts, and how long until something happens.