The Full SLR Equation: Meaning in Context

Interpreting the Full Regression Equation

In the previous lesson, you worked with individual coefficients. Now, let's put them together. The simple linear regression equation, $\hat{Y}={\hat{\beta }}_0+{\hat{\beta }}_{1}X$, is fundamentally a formula for making predictions. Given a value for the predictor variable $X$, you can calculate the model's best guess for the response variable, $\hat{Y}$.

An exam-style interpretation of the full equation must clearly state this predictive purpose and combine the meaning of both the intercept and the slope. The following reading provides an excellent, concise template for this.

The Exam-Style Template

Based on the principles from the reading, here is a robust template for interpreting the full regression equation:

The Template:
"The regression model predicts [Y variable] based on [X variable]. The model estimates a baseline [Y value] of [${\hat{\beta }}_0$ value] when [X value] is 0, and for each one-[unit] increase in [X variable], the predicted [Y variable] is expected to [increase/decrease] by [${\hat{\beta }}_1$ value]."

Let's apply this to the sales data from our last lesson, where the equation was:
Sales = 0.8031 + 0.6929 * No_Ads

Interpretation:
"The regression model predicts sales revenue based on the number of ads run. The model estimates a baseline sales revenue of 0.8031 units when zero ads are run. For each additional ad, the predicted sales revenue is expected to increase by 0.6929 units."

This two-sentence structure clearly defines the model's purpose and integrates the specific meaning of both coefficients in context.

The Prediction Error: Defining Residuals

Now that we know how to use the equation to get a predicted value ($\hat{y}$), the next logical question is: how good is that prediction? To answer this for a single data point, we calculate the residual.

A residual is simply the difference between the actual, observed value ($y$) and the value predicted by the model ($\hat{y}$). It is the prediction error for a single case.

The formal definition is straightforward:

The following reading explains this concept clearly.

Interpreting a Residual

The sign and size of the residual tell you everything about the prediction error for that specific data point:

• If $e>0$: This means $y>\hat{y}$. The actual value was higher than the prediction. The model under-predicted.
• If $e<0$: This means $y<\hat{y}$. The actual value was lower than the prediction. The model over-predicted.
• If $e=0$: This means $y=\hat{y}$. The model's prediction was perfect.

The larger the absolute value of the residual, the larger the prediction error.

Practice Drill

Let's put this together with an exam-style question. An analysis was run on 10 WNBA players to predict weight based on height.

Given Information:

• The regression equation is: Weight = -304 + 6.57 * Height
• One player is 75 inches tall and has an actual weight of 184 lbs.

Your Task:

1. Using the template, write a 2-line interpretation of the full regression equation in context.
2. For the player who is 75 inches tall:
   a. Calculate her predicted weight using the equation.
   b. Calculate her residual and interpret what it means.

Solutions

1. Interpretation of the Equation:

   • Answer: "The model predicts a player's weight (in lbs) based on their height (in inches). The model estimates a baseline weight of -304 lbs when height is 0 (an impractical extrapolation), and for each additional inch of height, the predicted weight is expected to increase by 6.57 lbs."
   • Note: Acknowledging that the intercept is a nonsensical extrapolation shows a higher level of understanding, which is great for impressing a strict professor.

2. Prediction and Residual:

   • a. Predicted Weight:
     $\hat{y}=-304+6.57\times 75=-304+492.75=188.75\text{ lbs}$
   • b. Residual:
     $e=y-\hat{y}=184-188.75=-4.75\text{ lbs}$
   • Interpretation: The residual is -4.75 lbs. This means the model over-predicted this player's weight by 4.75 lbs.

Conclusion

In this lesson, you've learned how to synthesize the meanings of the intercept and slope to interpret the full regression equation as a predictive tool. You also now understand that a residual measures the prediction error for an individual observation.

Key takeaways for your exam:

• Full Equation Interpretation: Combine the baseline prediction (intercept) with the rate of change (slope) into a single, contextual statement.
• Residual Definition: $e=y-\hat{y}$. It is the observed value minus the predicted value.
• Residual Interpretation: A positive residual means the model under-predicted; a negative residual means it over-predicted.

We have now covered all the foundational concepts for interpreting a simple linear regression model. In our next lesson, we will move on to assessing the overall quality of the model by looking at metrics that summarize all the residuals at once: R-squared and the Residual Standard Error.