Interpreting and Predicting From a Linear Model Revision Notes for VCE SSCE General Mathematics

Interpreting and Predicting From a Linear Model

Understanding slope and intercept in context

When you work with linear regression, you create an equation in the form $y = a + bx$ . This equation doesn't just give you numbers - it tells a story about the relationship between your variables. Learning to interpret the slope and intercept helps you understand what the data is really showing you.

The slope

The slope (represented by $b$ in the equation) shows you the rate of change. Specifically, it tells you how much the response variable changes, on average, when the explanatory variable increases by one unit.

For example, if you have the equation:

$\text{mark} = 30.8 + 1.62 \times \text{time}$

The slope is 1.62. This means that for each additional hour a student studies, their exam mark increases by an average of 1.62 percentage points.

The intercept

The intercept (represented by $a$ in the equation) tells you the expected value of the response variable when the explanatory variable equals zero.

Using the same equation above, the intercept is 30.8. This suggests that a student who does not study at all (time = 0) would obtain an average mark of 30.8%.

chatImportant

The intercept may not always make practical sense, especially when zero is not within your data range. For instance, if you're studying the relationship between height and weight, an intercept suggesting weight when height is zero centimetres wouldn't be meaningful.

Interpreting slope and intercept

infoNote

For any regression line $y = a + bx$ :

The slope $(b)$ represents the average change in the response variable $(y)$ for each one-unit change in the explanatory variable $(x)$
The intercept $(a)$ represents the average value of the response variable $(y)$ when the explanatory variable $(x)$ equals zero

lightbulbExample

Worked Example: Interpreting slope and intercept

A regression line models the association between study time (in hours) and exam marks (as percentages). The equation is:

$\text{mark} = 30.8 + 1.62 \times \text{time}$

Part a: Intercept

The intercept is 30.8.

This tells us that, on average, students who spend no time studying for the examination will obtain a mark of 30.8%.

Part b: Slope

The slope is 1.62.

This tells us that, on average, students' marks increase by 1.62 percentage points for each extra hour of study.

chatImportant

Exam tip: When interpreting slope and intercept, always include:

The word "average" or "on average"
The context (what the variables represent)
The units of measurement
Whether values are increasing or decreasing

Making predictions using a linear model

Once you have a regression equation, you can use it to make predictions by substituting values for the explanatory variable. However, you need to be careful about whether you're making predictions within or outside your data range.

The study time example

Consider this data showing study time and exam marks for 10 students:

If we fit a linear model using the least squares method, we get the equation:

$\text{mark} = 30.8 + 1.62 \times \text{time}$

Using this equation, we can predict marks for different study times:

For 0 hours: $\text{mark} = 30.8 + 1.62 \times 0 = 31\%$
For 8 hours: $\text{mark} = 30.8 + 1.62 \times 8 = 44\%$
For 12 hours: $\text{mark} = 30.8 + 1.62 \times 12 = 50\%$
For 30 hours: $\text{mark} = 30.8 + 1.62 \times 30 = 79\%$
For 80 hours: $\text{mark} = 30.8 + 1.62 \times 80 = 160\%$

chatImportant

The last prediction shows a major problem - it's impossible to score 160% on an exam! This highlights why we need to understand the difference between interpolation and extrapolation.

Interpolation and extrapolation

The reliability of predictions from a linear model depends on whether you're predicting within or outside the range of your original data.

Interpolation

Interpolation means making predictions within the range of values of the explanatory variable that you used to create the regression equation.

In the study time example, students studied between 1 and 36 hours. Any prediction for study time within this range (say, 8, 12, or 30 hours) is interpolation. These predictions are generally reliable because you're working within the data range where you know the linear relationship holds.

infoNote

Mnemonic: IN-terpolation = IN the data range = reliable

Extrapolation

Extrapolation means making predictions outside the range of values of the explanatory variable.

In the study time example, predicting marks for 80 hours of study is extrapolation because it's well beyond the maximum study time of 36 hours in your data. These predictions are generally unreliable because you don't know if the linear relationship continues outside your data range.

infoNote

Mnemonic: EX-trapolation = EX-ternal to data range = unreliable

Key differences:

Interpolation	Extrapolation
Predictions within the data range	Predictions outside the data range
Generally reliable	Generally unreliable
Safe to use	Use with caution
Linear relationship is known to hold	Linear relationship may not continue

Why extrapolation is risky

chatImportant

When you extrapolate, several problems can occur:

The relationship may no longer be linear outside your data range
You may get impossible values (like 160% or negative weights)
You're making assumptions about data you haven't observed
The model may break down completely

infoNote

Exam tip: Always check whether a prediction involves interpolation or extrapolation. If extrapolating, note that the prediction is less reliable.

Worked example: Making predictions

lightbulbExample

Worked Example: Making predictions

The equation relating weights (in kg) and heights (in cm) of students whose heights ranged from 163 cm to 190 cm is:

$\text{weight} = -40 + 0.60 \times \text{height}$

Part a: Predict weight for height = 170 cm

Substituting 170 into the equation:

\begin{aligned} \text{weight} &= -40 + 0.60 \times 170 \\ \text{weight} &= -40 + 102 \\ \text{weight} &= 62 \text{ kg} \end{aligned}

This is interpolation because 170 cm is within the range of heights (163 to 190 cm). The prediction is reliable.

Part b: Predict weight for height = 65 cm

Substituting 65 into the equation:

\begin{aligned} \text{weight} &= -40 + 0.60 \times 65 \\ \text{weight} &= -40 + 39 \\ \text{weight} &= -1.0 \text{ kg} \end{aligned}

This is impossible - a person cannot have negative weight!

This is extrapolation because 65 cm is well outside the range of heights (163 to 190 cm). The prediction is unreliable and, in this case, completely meaningless.

Steps for making predictions

infoNote

Follow these steps when making predictions:

Identify the regression equation
Note the range of the explanatory variable in your data
Substitute the value into the equation
Calculate the predicted value
Determine if you're interpolating or extrapolating
Comment on the reliability of your prediction

bookmarkSummary

Key Points to Remember:

The slope tells you the average change in the response variable for each one-unit change in the explanatory variable
The intercept tells you the expected value of the response variable when the explanatory variable equals zero (though this may not always be meaningful)
Interpolation (predicting within your data range) gives reliable predictions
Extrapolation (predicting outside your data range) gives unreliable predictions and can produce impossible results
Always check whether your prediction is interpolation or extrapolation, and comment on its reliability

Interpreting and Predicting From a Linear Model (VCE SSCE General Mathematics): Revision Notes