Writing the Equation With Variable Names

Make the Regression Equation Say What It Predicts

In “Reading the Equation of a Regression Line,” the fitted line was written as \(\hat{y}=a+bx\). That form is compact, but it uses symbols that do not identify the variables on their own. When a problem gives variables such as “daily screen time” and “battery drain,” writing the equation with those names can make the predictor and the predicted response much easier to recognize.

The task is a change in notation, not a different regression model. The predictor still takes the place of \(x\), the predicted response still takes the place of \(\hat{y}\), the intercept remains the constant term, and the slope still multiplies the predictor. This is the same line discussed in “Finding the Equation With LinReg(a+bx)” and “Reading Regression Computer Output,” now written with variable names in place of generic symbols.

For example, if a line predicts travel time from route distance, you could write \(\widehat{\text{travel time}}=a+b(\text{route distance})\). The hat indicates that the left side is a predicted value, not an observed travel time. The name in parentheses identifies the predictor whose value is used to obtain that prediction.

The parentheses are useful when a variable name contains multiple words. They show that the slope is multiplied by the predictor value as a whole. They are not essential if the notation is otherwise clear, but they help avoid reading adjacent words or symbols as part of the coefficient.

A Consistent Translation from Generic Symbols

Before substituting coefficients, name the variables in their roles. The response is the variable the line predicts; the predictor is the explanatory variable used as input. These roles were central in “Reading Regression Computer Output.” A variable’s position in a written equation should reflect its role, not simply the order in which the problem mentions it.

In symbols, the translation is:

The words “predicted response” and “predictor” in this general pattern are placeholders. Replace them with the actual variable names from the situation. For instance, if the response is weekly water use and the predictor is the number of hot days, use those names—not the generic placeholders—in the final equation.

Keep the hat on the response name, not on the predictor name. The predictor is an input value; the left side is the predicted response. This distinction also keeps the equation separate from an observed pair of measurements. An observed response is not necessarily equal to its predicted value; as covered in “Predicted Change Versus Actual Change,” observations can differ from the line’s predictions.

Worked Examples

Choose Names That Make the Roles Clear

A descriptive name should be specific enough to identify the measured quantity. “Time” may be unclear if the situation includes several kinds of time. “Customer wait time” is more informative. Similarly, if a problem distinguishes distance ridden from distance remaining, use the name that exactly matches the predictor used to fit the line.

Include units in the surrounding explanation when the variable name alone does not show them. For example, you might define the response as “daily water use, in liters” and the predictor as “outdoor temperature, in degrees Celsius,” then write the equation using the shorter names. Units are not a substitute for identifying the variables, but they help make the equation unambiguous.

Long names can make equations visually crowded. It is acceptable to define short labels first, as long as the link to the real variables is clear. For example, define \(W\) as daily water use in liters and \(T\) as outdoor temperature in degrees Celsius, then write \(\widehat{W}=18.4+2.7T\). This still uses letters, but the definitions make their meanings explicit. If the instruction specifically asks for descriptive variable names, show the names in the final equation rather than relying only on the abbreviations.

Common Mistakes and AP Exam Tips

• Putting the predictor on the left. The left side is the predicted response, not the predictor. Identify the variable the model predicts before writing the equation.
• Leaving off the hat. A hat distinguishes a predicted response from an observed response. Write \(\widehat{\text{wait time}}\), for example, when the equation gives predicted wait time.
• Putting a hat on the predictor. The predictor is the input to the line. Keep the hat on the predicted response name, not on the name in parentheses.
• Writing the slope as a constant. The slope multiplies the predictor. An equation such as \(\widehat{\text{water use}}=18.4+2.7\) leaves out the predictor and is not the fitted line.
• Reversing the intercept and slope. The intercept is the constant term; the slope is attached to the predictor. Use the coefficient roles established in “Reading Regression Computer Output,” not the order of terms in a table.
• Dropping or changing the sign. Copy a negative slope as negative. Writing \(+1.35(\text{number of open counters})\) instead of \(-1.35(\text{number of open counters})\) changes the fitted equation.
• Using names that do not match the question. A generic label such as “amount” can leave the response unclear. Name the actual measured quantity and, where useful, state its units.

A full-credit response makes the roles visible. For example: “Customer wait time is the response, and number of open counters is the predictor. Therefore, the fitted equation is \(\widehat{\text{customer wait time}}=11.2-1.35(\text{number of open counters})\), with wait time measured in minutes.” That statement identifies the variables and writes the equation without requiring the reader to guess what \(x\) and \(y\) represent.