01 / 20

Linear Regression
Interpretation

Learn to read, interpret, and apply the least-squares regression line \( \hat{y} = a + bx \) using technology-generated output.

College Algebra

What Is Linear Regression?

Linear regression is a statistical method that finds the best-fitting straight line through a set of data points. The line minimizes the total squared vertical distances (residuals) between each point and the line.

Key Vocabulary

Slope \( b \)
The predicted change in \( y \) for each one-unit increase in \( x \).
\( y \)-Intercept \( a \)
The predicted value of \( y \) when \( x = 0 \).
\( r \) (Correlation)
Measures the strength and direction of the linear relationship: \( -1 \leq r \leq 1 \).
\( r^2 \) (Coefficient of Determination)
The proportion of variance in \( y \) explained by the linear model.

Visualizing the Best-Fit Line

Study Hours (x) Exam Score (y) 0 2 4 6 8 10 50 60 70 80 90 residual Data Best fit

The Regression Equation

$$ \hat{y} = a + bx $$

\( b \) : Slope
"For every one-unit increase in \( x \), \( \hat{y} \) changes by \( b \) units."
\( a \) : Intercept
"When \( x = 0 \), the predicted value of \( y \) is \( a \)."

Understanding \( r \) and \( r^2 \)

1
Correlation \( r \): measures direction and strength. Positive \( r \) means both variables increase together; negative means one decreases as the other increases.
2
Coefficient of determination \( r^2 \): the fraction of variability in \( y \) that is explained by the regression on \( x \). For example, \( r^2 = 0.85 \) means 85% of the variation in \( y \) is explained by the model.
3
Key relationship: \( r^2 = (r)^2 \). If \( r = -0.9 \), then \( r^2 = 0.81 \). The sign of \( r \) matches the sign of \( b \).

Correlation Strength Scale

-1 -0.5 0 0.5 1 Strong negative No linear relation Strong positive r = 0.92

Correlation vs Causation

Correlation
Two variables move together. We can describe the pattern and make predictions within the data range.
Causation
One variable directly causes a change in the other. Requires a controlled experiment to establish.

How to Interpret Tech Output

1
Identify \( a \) and \( b \): locate the intercept and slope in the calculator or software output. Write the equation \( \hat{y} = a + bx \).
2
Interpret slope in context: "For each additional [unit of \( x \)], the predicted [response] changes by \( b \) [units of \( y \)]."
3
Interpret intercept in context: "When \( x = 0 \), the predicted [response] is \( a \)." Decide if \( x = 0 \) is meaningful.
4
Read \( r \) and \( r^2 \): state the direction, strength, and percent of variation explained.
5
Make predictions (carefully): substitute an \( x \)-value into the equation. Avoid extrapolation outside the data range.

Example 1: Study Hours vs Exam Scores

A calculator gives this regression output for 10 students' study hours (\( x \)) and exam scores (\( y \)):

Example 2: Making a Prediction

Using \( \hat{y} = 51.07 + 3.65x \), predict the exam score for a student who studies 6.5 hours.

1
Substitute: \( \hat{y} = 51.07 + 3.65(6.5) \)
2
Multiply: \( 3.65 \times 6.5 = 23.725 \)
3
Add: \( 51.07 + 23.725 = 74.795 \)

Example 3: Negative Slope

A researcher studies car age (\( x \), years) versus resale value (\( y \), dollars). Technology output:

Interpolation vs Extrapolation

Data range (interpolation safe) Extrapolation (risky) Safe Risky x min x max
Interpolation
Predicting within the range of observed \( x \)-values. Generally reliable.
Extrapolation
Predicting beyond the data range. The linear pattern may not continue, making predictions unreliable.

Common Mistakes

✕ Claiming Causation
Saying "studying causes higher scores" from regression alone. Regression shows association, not causation.
✕ Ignoring Context
Reporting "slope = 3.65" without units or context. Always include variable names and units.
✕ Extrapolating Blindly
Predicting for \( x = 50 \) hours when data only goes to 10. The model does not apply outside its range.

Summary Table

Component Symbol What It Tells You
Slope \( b \) Predicted change in \( y \) per one-unit increase in \( x \)
\( y \)-Intercept \( a \) Predicted \( y \) when \( x = 0 \)
Correlation \( r \) Direction and strength of linear relationship
Coefficient of Determination \( r^2 \) Proportion of variance in \( y \) explained by model
Residual \( y - \hat{y} \) Difference between observed and predicted value

Question 1 of 5

A regression equation is \( \hat{y} = 12.4 + 2.8x \), where \( x \) is hours of exercise per week and \( y \) is resting heart rate decrease (bpm). What does the slope mean?

Question 2 of 5

If \( r^2 = 0.81 \) for a regression model, which of the following is the best interpretation?

Question 3 of 5

A regression model for tree age (\( x \), years) vs height (\( y \), feet) gives \( \hat{y} = 2.1 + 1.4x \) for trees aged 5 to 30. What is the concern with predicting the height of a 50-year-old tree?

Question 4 of 5

Technology output shows \( r = -0.85 \) for the relationship between outdoor temperature (\( x \)) and heating cost (\( y \)). What does this tell us?

Question 5 of 5

Given \( \hat{y} = 5.2 + 0.75x \), what is the predicted value of \( y \) when \( x = 8 \)?