In AP Precalculus, regression is a technology-based technique that fits a function model (linear, quadratic, cubic, quartic, or exponential) to a data set by finding the best-fitting curve of a specified type, which you then validate by checking whether its residual plot shows no pattern.
Regression is how you turn a messy table of data into a clean function you can actually use. You tell your calculator what type of function to fit (linear, quadratic, cubic, quartic, or exponential), and it finds the version of that function that comes closest to the data points overall. The output is a model, an equation like f(x) = 2.3x² - 1.1x + 4.7 that approximates the real-world relationship.
Here's the part the AP exam cares about most. The calculator will always give you an answer, even if you picked a terrible function type. Your job is the thinking before and after the button-press. Before, you choose the function type based on the pattern in the data and the context (per 1.14.A). After, you validate the model by graphing the residuals, the differences between actual and predicted values. If the residual plot looks like random scatter with no pattern, the model is appropriate. If it shows a U-shape, fan, or other pattern, your model type is wrong (per 2.6.B).
Regression lives in two places in the CED. In Unit 1, Topic 1.14 (Function Model Construction and Application), learning objectives 1.14.A and 1.14.B say you can use regression to fit polynomial models, including linear, quadratic, cubic, and quartic regressions, and then use those models to answer contextual questions (1.14.D). In Unit 2, Topic 2.6 (Competing Function Model Validation), objectives 2.6.A and 2.6.B push further. There you construct linear, quadratic, AND exponential models for the same data, compare them, and use residual plots to decide which one is actually justified. Regression is also the bridge skill for the calculator-active part of the exam, since the CED expects you to construct these models with technology, not by hand.
Keep studying AP® Precalculus Unit 1
Residual plot (Unit 2)
The residual plot is the report card for your regression. A patternless scatter of residuals means the model fits; a U-shape or fan means you chose the wrong function type. You can't validate a regression without it.
Quadratic regression (Units 1-2)
Quadratic regression is the move when a linear regression's residuals form a U-shape. The U is the curvature your line couldn't capture, and a degree-2 model soaks it up.
Cubic regression (Unit 1)
When data changes direction more than once, a quadratic can't keep up. Cubic (and quartic) regression extends the same fit-a-polynomial idea to higher degrees under 1.14.B.
Inversely proportional relationships (Unit 1)
Not every data set wants a polynomial. Per 1.14.C, quantities that are inversely proportional (like gravitational force vs. squared distance) call for a rational function model instead. Context tells you which family to fit before you ever hit the regression button.
Regression shows up two ways. On multiple choice, the classic stem hands you a residual plot and asks you to judge the model. A U-shaped residual pattern means the model is inappropriate and you should try a higher-degree or different-family model (a U after linear regression points to quadratic). A fan shape means the error grows as x grows. Random scatter around zero means the model is justified. You may also be asked to compare three competing models (linear, quadratic, exponential) for the same data and pick the one whose residuals are patternless. On the FRQ side, modeling questions give you a table of values, like the 2025 FRQ Q1 table for a decreasing function, and expect you to construct the right model type from the data's pattern, then use it to predict values or rates of change with correct units. Know your calculator's regression commands cold for the calculator-active sections.
Regression builds the model; the residual plot judges it. Regression is the technology process that produces the best-fitting curve of a chosen type. The residual plot graphs the leftover errors (actual minus predicted) so you can see whether that curve actually fits. A regression can always be computed, but only a patternless residual plot makes it appropriate.
Regression uses technology to fit a function of a specified type (linear, quadratic, cubic, quartic, or exponential) to a data set.
You choose the function type based on the pattern in the data and the context, then let the calculator find the best-fitting version of that type.
A regression model is justified only if its residual plot shows no pattern; random scatter around zero means the model is appropriate.
A U-shaped residual plot after a linear regression signals curvature, so a quadratic (or higher-degree) model is the better next try.
The residual, actual value minus predicted value, is the model's error, and context decides whether underestimating or overestimating is more acceptable.
Once a model is built, you use it to predict values, rates of change, and average rates of change, always with appropriate units from the context.
Regression is a technology-based technique that fits a function model to a data set by finding the best-fitting curve of a type you specify. The CED includes linear, quadratic, cubic, quartic, and exponential regressions, covered in Topics 1.14 and 2.6.
No. Your calculator will compute a regression for any function type you choose, even a bad one. The model is only justified if its residual plot shows no pattern, which is exactly what learning objective 2.6.B asks you to check.
The regression creates the model; the residual plot tests it. Residuals are the differences between actual data values and the model's predictions, and graphing them reveals whether the model type fits or needs to change.
It means your model missed curvature in the data, so the model is not appropriate. If a linear regression produces a U-shaped residual plot, a quadratic regression is the standard next move.
Linear, quadratic, and exponential for comparing competing models in Topic 2.6, plus cubic and quartic polynomial regressions in Topic 1.14. You run all of them with your calculator, so know its regression menu well before exam day.
Connect this key term to the AP exam workflow: review the course, practice questions, and check related study tools.
Review units, study guides, and course resources.
Check this vocabulary in multiple-choice context.
Apply key concepts in written AP responses.
Estimate the exam score you are working toward.
Review the highest-yield facts before practice.
Put the full course together before test day.