Polynomial regression shows up in AP Statistics, intro data science courses, and the first week of any machine learning class — and most students hit it cold, with a textbook that buries the core idea under pages of theory before getting to anything useful.

This TLDR primer cuts straight to what matters. You'll learn how polynomial regression extends linear models to fit curved data, how the coefficients are actually computed using least squares and the normal equations, and how to tell whether a fit is genuinely good or just memorizing noise. The guide then tackles the concept that trips up almost every student encountering predictive modeling for the first time: the **bias-variance tradeoff**. Cranking up the polynomial degree always lowers error on your training data — and almost always hurts you on new data. This book shows you exactly why, and gives you concrete tools (train/test splits, k-fold cross-validation) to choose the right degree without guessing.

The final section covers the honest limits of polynomial models: why extrapolation fails, how multicollinearity between powers of *x* inflates uncertainty, and when a spline or a different model family is the smarter call.

Written for high school students in statistics or pre-calculus, early college students in data science or applied math, and anyone who wants to walk into a regression unit with real understanding instead of memorized formulas. Concise and to the point — no filler, no detours.

If polynomial regression is on your syllabus, start here.

• Set up a polynomial regression as a linear least-squares problem and solve it for small degrees by hand and conceptually for larger ones.
• Interpret R-squared, residuals, and residual plots to judge whether a polynomial fit is appropriate.
• Recognize overfitting and the bias-variance tradeoff, and use train/test splits or cross-validation to pick a degree.
• Understand the limits of polynomial models — extrapolation failure, multicollinearity of powers, and when to prefer splines or other models.

1. 1. From Lines to Curves: What Polynomial Regression Is
   Introduces polynomial regression as an extension of linear regression that fits curved relationships while still being linear in its coefficients.
2. 2. Fitting the Curve: Least Squares and the Normal Equations
   Walks through how the coefficients are actually computed by minimizing squared residuals, with a worked quadratic fit on a small dataset.
3. 3. Reading the Fit: R-squared, Residual Plots, and What 'Good' Means
   Shows how to evaluate a polynomial fit using R-squared, adjusted R-squared, and visual diagnostics, and warns against common interpretation traps.
4. 4. Overfitting and the Bias-Variance Tradeoff
   Explains why cranking up the degree always lowers training error but eventually wrecks predictions, and frames the tradeoff that drives model selection.
5. 5. Choosing the Degree: Train/Test Splits and Cross-Validation
   Gives concrete procedures for picking the right polynomial degree using held-out data and k-fold cross-validation.
6. 6. Limits and Alternatives: When Not to Use a Polynomial
   Covers extrapolation failure, multicollinearity between powers of x, and points toward splines, logistic regression, and other models for cases polynomials handle poorly.