Part 7: Relationships and Prediction
Regression: the statistical technique you'll use most often in the real world.
If you go into business, you'll use regression to predict sales. If you go into medicine, you'll use regression to model risk factors. If you go into social science, you'll use regression to control for confounding variables. If you go into data science or AI, regression is the foundation everything else is built on. It's not an exaggeration to say that regression is the single most widely used statistical method on the planet.
And the core idea is surprisingly intuitive: given what I know about X, what can I predict about Y?
That question — the relationship between variables — is what this entire part of the book is about.
Chapter 22 starts with correlation and simple linear regression. You'll learn to measure the strength of a linear relationship using the correlation coefficient, fit a line through a scatterplot using least squares, and interpret the slope and intercept in plain language. You'll also spend real time on the distinction between correlation and causation — the single most important critical thinking habit in all of statistics. Just because two things are related doesn't mean one causes the other, and understanding why is essential for navigating a world full of misleading claims.
Chapter 23 levels up to multiple regression, where the outcome variable depends on more than one predictor. This is where statistics gets genuinely powerful. Instead of asking "does education predict income?" you can ask "does education predict income after accounting for experience, industry, and geographic location?" The phrase "holding other variables constant" is the conceptual key, and once you understand it, you'll read research papers, news stories, and policy arguments with much sharper eyes.
Chapter 24 handles a specific but incredibly common situation: what if the thing you're trying to predict is a yes-or-no outcome? Will the customer buy or not? Will the patient survive or not? Will the email be spam or not? Linear regression doesn't work for binary outcomes, but logistic regression does. This chapter also connects directly to AI and machine learning, because logistic regression is literally the simplest classification algorithm — the one every machine learning course starts with.
Chapter 25 brings it all together with data communication. You've spent the entire book learning to analyze data. Now you'll learn to present it — to tell clear, honest, compelling stories with your results. You'll learn visualization best practices, how to write for non-technical audiences, and how to spot (and avoid) misleading graphs. Because the best analysis in the world is useless if nobody understands it.
By the end of Part 7, you'll have the skills to model relationships, make predictions, and communicate your findings. That's not just a statistics education — that's a career-ready skill set.