A residual plot is a graph of a regression model's residuals (actual value minus predicted value) plotted against the input variable. In AP Precalculus, a residual plot with no pattern justifies the model as appropriate; a curved or U-shaped pattern means you need a different model.
A residual is the gap between what your model predicts and what the data actually says, so residual = actual − predicted. A residual plot takes all those gaps and graphs them against the independent variable. It's basically a report card for your regression, where the original scatterplot gets flattened out and only the errors remain.
Here's the one rule the CED cares about (EK under 2.6.B): if the residuals look like random scatter with no pattern, the model is justified as appropriate for the data. If the residuals form a pattern, especially a U-shape or curve, the model is systematically missing something. For example, fitting a line to data that's actually quadratic leaves a curved residual plot because the line underestimates at the ends and overestimates in the middle (or vice versa). The residual plot exposes structure your model failed to capture.
Residual plots live in Topic 2.6, Competing Function Model Validation, in Unit 2 (Exponential and Logarithmic Functions). They directly support learning objective 2.6.B, validating a model constructed from a data set, and they're the tool you use after 2.6.A, where you build linear, quadratic, and exponential models from the same data. Data with a slightly changing rate of change can plausibly be modeled by all three function types, so just eyeballing the scatterplot isn't enough. The residual plot is how AP Precalculus has you settle the competition. Random scatter wins; visible pattern loses. The CED also notes that residuals tell you about the model's error, and that in context, an underestimate or overestimate on a given interval might matter (predicting too little medicine is a different problem than predicting too much).
Keep studying AP® Precalculus Unit 2
Regression (Unit 2)
Regression builds the model; the residual plot judges it. Every residual plot starts with a regression, because you can't compute predicted-versus-actual gaps without a prediction. Think of them as two halves of one workflow in Topic 2.6.
Competing Linear, Quadratic, and Exponential Models (Unit 2)
Under LO 2.6.A, the same data set can get fit with a linear, quadratic, and exponential model. The classic exam setup gives you three residual plots and asks which model wins. The one showing random scatter is the appropriate model, no matter how 'fancy' the other equations look.
Error in a Model: Underestimates and Overestimates (Unit 2)
Each point on a residual plot is an error. A positive residual means the model underestimated the actual value; a negative residual means it overestimated. The CED points out that context decides which kind of error is more acceptable, like preferring an overestimate when predicting how much vaccine to stock.
This shows up in multiple-choice questions that hand you residual plot descriptions and ask you to pick the appropriate model or interpret a pattern. Two stems to expect: (1) three models are fit to the same data, the linear residual plot shows a clear curved pattern, the quadratic shows random scatter, and you choose the quadratic; (2) a residual plot shows a distinct U-shape, and you conclude the model is NOT appropriate because the pattern reveals systematic error. Watch out for the twist where even a quadratic model can produce a U-shaped residual plot, which means that quadratic is also wrong for that data. The rule is about the pattern, not the model type. You may also need to interpret a single residual's sign as an underestimate or overestimate in context.
A scatterplot graphs the actual data values; a residual plot graphs only the errors after a model has been fit. The scatterplot of good data can show a strong curve or trend, and that's fine. It's the residual plot that should show NO trend. Students mix these up and panic when the data scatterplot has a shape. Shape in the data is expected; shape in the residuals is the red flag.
A residual is the actual value minus the predicted value, and a residual plot graphs these errors against the independent variable.
A residual plot with no pattern (random scatter) justifies the model as appropriate for the data set, per the essential knowledge for LO 2.6.B.
A U-shape or curved pattern in a residual plot means the model is systematically wrong, even if the model itself is quadratic or exponential.
When comparing competing linear, quadratic, and exponential models, the appropriate one is the model whose residual plot shows random scatter.
A positive residual means the model underestimated the actual value, and a negative residual means it overestimated; context decides which error is more acceptable.
It's a graph of a regression model's residuals (actual minus predicted values) plotted against the input variable. In Topic 2.6, you use it to validate whether a linear, quadratic, or exponential model is appropriate for a data set.
No, the opposite. A U-shape (or any clear pattern) means the model is systematically missing the data's behavior and is not appropriate. You want random, patternless scatter in the residuals.
A scatterplot shows the actual data, which can have any shape, like a curve for exponential growth. A residual plot shows only the leftover errors after fitting a model, and that one should look like random noise if the model is appropriate.
Yes. If a quadratic regression's residuals form a clear U-shape, the quadratic model is not appropriate for that data either. The no-pattern rule applies to every model type, not just linear ones.
The actual value was higher than the model predicted, so the model underestimated. A negative residual means the model overestimated. The CED notes that depending on context, one kind of error may be more acceptable than the other.
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.