Unit 10 · Lesson 10.5

10.5Residuals

Calculate and interpret residuals — the difference between actual and predicted values. Use residual plots to evaluate whether a linear model is a good fit for the data.

Why This Matters

Residuals measure how well a model fits the data — a concept central to regression analysis in AP Statistics and data science. Understanding residuals helps you evaluate whether a linear model is appropriate or if a different model is needed.

Workbook

Lesson, vocabulary, worked examples, and practice problems.

Essential Question

How do residuals help us evaluate how well a linear model fits a set of data?

Lesson Overview

A residual is the difference between an actual data value and the value predicted by the line of best fit: residual = y − ŷ (actual minus predicted). A positive residual means the actual value is above the line (the model underestimated). A negative residual means the actual value is below the line (the model overestimated). A residual plot graphs the residuals against the x-values. If the residuals are randomly scattered with no pattern, the linear model is a good fit. If the residuals show a curved or systematic pattern, a nonlinear model may be more appropriate.

Residual Formula

Residual = Actual − Predicted = y − ŷ

Positive residual (y > ŷ)

Point is above the line. Model underestimates.

Zero residual (y = ŷ)

Point is exactly on the line. Perfect prediction.

Negative residual (y < ŷ)

Point is below the line. Model overestimates.

Residuals — Actual vs. Predicted

2468ŷ = x+1

Blue dots = actual data points (y)

Red dots = predicted values (ŷ) on the line

Green dashed = positive residual (above line)

Actual > Predicted → residual > 0

Purple dashed = negative residual (below line)

Actual < Predicted → residual < 0

Residual Plot — What to Look For

Good fit — Random scatter

Residuals are randomly scattered above and below the x-axis with no pattern. The linear model is appropriate.

+ · · − · + − · + ·
(random, no trend)

Poor fit — Pattern visible

Residuals show a curve or systematic pattern. The linear model may not be the best fit — a nonlinear model might be better.

+ + + − − − + + +
(curved pattern)

Worked Examples

Example 1

The line of best fit is ŷ = 2x + 3. A data point is (4, 14). Calculate the residual.

Predicted: ŷ = 2(4) + 3 = 11

Residual = y − ŷ = 14 − 11 = 3

Answer:Residual = 3 (positive — point is above the line)
Example 2

The line of best fit is ŷ = 5x + 10. A data point is (3, 20). Calculate the residual and interpret it.

Predicted: ŷ = 5(3) + 10 = 25

Residual = 20 − 25 = −5

Negative residual: the actual value (20) is below the predicted value (25). The model overestimates.

Answer:Residual = −5. The model overestimates by 5 units.
Example 3

For the line ŷ = 3x + 1, calculate residuals for the data points (1, 5), (2, 7), (3, 8), (4, 14).

x=1: ŷ=4, residual=5−4=1

x=2: ŷ=7, residual=7−7=0

x=3: ŷ=10, residual=8−10=−2

x=4: ŷ=13, residual=14−13=1

Answer:Residuals: 1, 0, −2, 1
Example 4

A residual plot shows residuals scattered randomly above and below the x-axis. What does this tell us about the linear model?

Random scatter in a residual plot means there is no systematic pattern.

This indicates the linear model is a good fit for the data.

Answer:The linear model is appropriate — residuals show no pattern.
Example 5

A residual plot shows a curved pattern (positive residuals at low x, negative in the middle, positive at high x). What does this suggest?

A curved pattern in the residual plot indicates the linear model is not the best fit.

The data may follow a nonlinear (curved) pattern.

A different model (e.g., quadratic) may be more appropriate.

Answer:The linear model is not a good fit. A nonlinear model may be more appropriate.

Guided Practice

Guided Practice Video: Residuals

Review how to calculate and interpret residuals — the difference between actual and predicted values — and how residual plots are used to assess the fit of a linear model before completing the guided problems below.

Video by Sang Real Math

Watch on YouTube ↗
Guided Problem 1

The line of best fit is ŷ = 4x + 2. A data point is (5, 24). Calculate the residual.

Hint: Predicted: ŷ = 4(5) + 2. Residual = actual − predicted.

Guided Problem 2

The line of best fit is ŷ = −2x + 30. A data point is (6, 15). Calculate the residual and interpret it.

Hint: Find ŷ first. Then residual = y − ŷ. Is the point above or below the line?

Guided Problem 3

For ŷ = 3x + 5, calculate residuals for (2, 12), (4, 17), (6, 23).

Hint: Calculate ŷ for each x, then subtract from the actual y.

Guided Problem 4

A residual plot shows residuals of +3, −1, +2, −3, +1, −2 for x = 1 through 6. Does the linear model appear to be a good fit? Explain.

Hint: Look for a pattern. Are the residuals randomly scattered or do they show a trend?

Guided Problem 5

Error Analysis: A student calculates the residual for the point (3, 10) on the line ŷ = 2x + 5 as residual = 11 − 10 = 1. Identify the error.

Hint: What is the correct formula for residual? Which value is actual and which is predicted?

Key Vocabulary

Residual

The difference between the actual y-value and the predicted ŷ-value: residual = y − ŷ.

Predicted Value (ŷ)

The y-value calculated by substituting x into the equation of the line of best fit.

Actual Value (y)

The observed data value from the scatter plot.

Positive Residual

Actual > Predicted. The data point is above the line. The model underestimates.

Negative Residual

Actual < Predicted. The data point is below the line. The model overestimates.

Residual Plot

A scatter plot of residuals (y-axis) vs. x-values (x-axis). Used to evaluate model fit.

Random Scatter

Residuals with no visible pattern — indicates a linear model is appropriate.

Least Squares Line

The line of best fit that minimizes the sum of the squared residuals.

Interactive Practice — 5 Questions

1

The line of best fit is ŷ = 2x + 4. A data point is (3, 12). What is the residual?

2

A residual of −5 means:

3

Which formula correctly calculates a residual?

4

A residual plot shows residuals randomly scattered above and below zero. This means:

5

ŷ = 4x + 1. Data point: (5, 18). The residual is:

Independent Practice

Independent Practice

1

ŷ = 3x + 1. Data point: (4, 15). Calculate the residual. Is the point above or below the line?

2

ŷ = 2x + 8. Calculate residuals for: (1, 11), (3, 14), (5, 18). Which point is farthest from the line?

3

A residual is −4. Is the data point above or below the line of best fit? Does the model overestimate or underestimate?

4

A residual plot shows a U-shaped pattern. What does this suggest about the linear model? What type of model might be better?

5

ŷ = 4x + 3. A data point has residual = +5. If x = 2, what is the actual y-value? Show your work.

⚠️

Common Mistakes

Calculating residual as predicted − actual instead of actual − predicted.

Residual = actual y − predicted ŷ. A positive residual means the point is above the line; negative means below.

Thinking a residual of zero means the model is perfect.

A residual of zero means the model predicted that one point exactly. The overall fit depends on all residuals together.

Confusing a residual plot with a scatter plot — plotting y vs. x instead of residuals vs. x.

A residual plot shows residuals (y − ŷ) on the y-axis and the x-values (or predicted values) on the x-axis.

Concluding a linear model is appropriate just because r is strong — without checking the residual plot.

Always check the residual plot. A random scatter of residuals confirms a linear model is appropriate. A curved pattern suggests otherwise.

💡

Math Tips

📌

Residual = actual − predicted = y − ŷ. Always subtract in this order.

📌

Positive residual: the actual point is ABOVE the line (model underestimates).

📌

Negative residual: the actual point is BELOW the line (model overestimates).

📌

A good residual plot shows points randomly scattered above and below zero with no pattern. A curved or fan-shaped pattern means the linear model is not ideal.

📌

The sum of all residuals for the least-squares line is always zero (or very close to it).