Logarithmic Regression on Calculator
Logarithmic Regression on Calculator, Logarithmic regression is a statistical technique used to model a relationship between two variables when the response changes rapidly at first and then gradually levels off as the predictor variable increases.
It is useful for analyzing nonlinear relationships where the rate of change decreases over time or as the predictor becomes larger.
The standard logarithmic regression model is:
y = a + b ln(x)
where:
- y = response variable
- x = predictor or explanatory variable
- a = intercept
- b = regression coefficient
- ln(x) = natural logarithm of x
Because the natural logarithm grows quickly for small values of x and more slowly as x increases, this model can describe relationships that increase or decrease rapidly before gradually leveling off.
When Should You Use Logarithmic Regression?
Logarithmic regression is useful when a scatterplot shows a pattern that changes rapidly initially and then becomes progressively flatter.
For example, suppose a measurement increases substantially when x changes from 1 to 2, but changes only slightly when x changes from 8 to 9. A logarithmic model may be appropriate for describing this type of relationship.
A logarithmic model can represent both increasing and decreasing relationships depending on the value of b.
Positive Logarithmic Relationship
If b > 0, y increases as x increases.
For example:
y = 10 + 5ln(x)
The curve increases rapidly at first and then gradually flattens.
Negative Logarithmic Relationship
If b < 0, y decreases as x increases.
For example:
y = 80 − 10ln(x)
The curve decreases rapidly at first and then becomes flatter.
Logarithmic Regression vs Exponential Regression
Logarithmic regression and exponential regression are different models.
Logarithmic regression
y = a + b ln(x)
The response changes rapidly initially and then tends to level off as x increases.
Exponential regression
y = abˣ
The response can increase or decrease at an increasingly rapid rate.
For example, logarithmic regression may be appropriate when a quantity shows diminishing increases, while exponential regression may be appropriate when growth or decay becomes increasingly rapid.
Therefore, you should examine a scatterplot before selecting the regression model.
Example Dataset
Suppose we have the following data:
| x | y |
|---|---|
| 1 | 76 |
| 2 | 55 |
| 3 | 43 |
| 4 | 35 |
| 5 | 30 |
| 6 | 26 |
| 7 | 23 |
| 8 | 21 |
The response variable decreases rapidly at first and then gradually levels off. This pattern can be modeled using logarithmic regression.
Step 1: Enter the Data on a TI-84
First, enter the x and y values into the calculator.
- Press STAT.
- Select 1:Edit.
- Enter the x-values into L1.
- Enter the corresponding y-values into L2.
For the example above:
L1: 1, 2, 3, 4, 5, 6, 7, 8
L2: 76, 55, 43, 35, 30, 26, 23, 21
Make sure that every x-value is paired with its corresponding y-value.
Step 2: Create a Scatterplot
Before performing the regression, it is useful to examine the relationship between x and y.
Press:
2nd → Y=
This opens STAT PLOT.
Choose Plot1 and turn it on.
Set the plot type to a scatterplot and select:
- Xlist = L1
- Ylist = L2
Then press ZOOM and select:
9: ZoomStat
The calculator will display the data points.
Look for a pattern where the data changes rapidly initially and then gradually flattens. This is one indication that logarithmic regression may be appropriate.
Step 3: Perform Logarithmic Regression
Once the data has been entered:
- Press STAT.
- Use the right arrow to select CALC.
- Scroll down to LnReg.
- Press ENTER.
The calculator will calculate the logarithmic regression model.
You can also enter:
LnReg L1,L2
and press ENTER.
The calculator will return the estimated regression coefficients.
Step 4: Interpret the Regression Equation
Suppose the TI-84 produces the following equation:
y = 76.21296 − 29.8634ln(x)
Here:
a = 76.21296
and:
b = −29.8634
Because b is negative, the model represents a decreasing logarithmic relationship.
The equation can be used to estimate y for different values of x.
Step 5: Make a Prediction
Suppose we want to estimate y when:
x = 8
Using:
y = 76.21296 − 29.8634ln(x)
substitute x = 8:
y = 76.21296 − 29.8634ln(8)
Since:
ln(8) ≈ 2.0794
the predicted value is approximately:
y ≈ 14.11
Therefore, the predicted response when x = 8 is approximately 14.11.
The exact prediction may vary slightly depending on the coefficients produced by the calculator from the dataset.
How to Graph the Logarithmic Regression Equation
You can graph the fitted regression curve together with the original data.
After running LnReg, enter the resulting equation into the Y= screen.
For example:
Y₁ = 76.21296 − 29.8634ln(X)
Then press GRAPH.
The graph allows you to visually compare the logarithmic regression curve with the observed data points.
A good-fitting model should generally pass close to the observations without showing a strong systematic pattern in the residuals.
How to Store the Regression Equation Automatically
The TI-84 can store the logarithmic regression equation directly in a function.
After selecting LnReg, you can specify Y1 as the storage location.
For example:
LnReg L1,L2,Y1
Then press ENTER.
The calculator will calculate the regression equation and store it in Y1.
You can then press GRAPH to display the fitted logarithmic curve.
How to Calculate R² for Logarithmic Regression
R², or the coefficient of determination, provides information about how much of the variation in the response is explained by the fitted regression model.
If regression diagnostics are enabled on the TI-84, additional regression statistics can be displayed.
To enable diagnostics:
2nd → 0 (CATALOG)
Find:
DiagnosticOn
Press ENTER, followed by ENTER again.
Then run LnReg again.
A higher R² generally indicates that the model fits the observed data more closely, but R² should not be used as the only criterion for selecting a regression model.
You should also consider the scatterplot, residuals, subject-matter knowledge, and whether the model makes sense for the data.
Important: x Must Be Positive
A standard logarithmic regression model uses:
ln(x)
The natural logarithm is defined only for positive values in the real-number system.
Therefore:
x > 0
is required for the standard logarithmic regression model.
For example:
- x = 1 → valid
- x = 5 → valid
- x = 0.5 → valid
- x = 0 → invalid
- x = −2 → invalid
If your dataset contains zero or negative x-values, you need to consider whether a transformation or a different regression model is appropriate.
How to Interpret the Coefficient b
In:
y = a + b ln(x)
the coefficient b determines the direction of the relationship.
If b > 0
As x increases, y tends to increase.
If b < 0
As x increases, y tends to decrease.
The effect of x is not constant. The amount by which y changes for a one-unit increase in x depends on the current value of x.
This is one of the important characteristics of logarithmic regression.
Common Applications of Logarithmic Regression
Logarithmic regression can be useful in situations involving diminishing changes.
Learning Curves
Performance may improve rapidly at first and then show progressively smaller improvements with additional experience or training.
Economics
Some relationships involving income, productivity, or other economic variables may show diminishing effects as a predictor increases.
Population and Environmental Studies
Certain relationships between environmental measurements and biological responses may show rapid initial changes followed by stabilization.
Business and Marketing
Customer response can sometimes increase quickly with an initial increase in an input and then show diminishing returns.
Scientific Measurements
Some physical and biological processes exhibit nonlinear relationships that can be approximated using logarithmic models.
Logarithmic Regression and Diminishing Returns
One of the most useful ways to understand a logarithmic model is through diminishing returns.
Consider:
y = 20 + 10ln(x)
When x increases from 1 to 2, the increase in y is relatively large.
When x increases from 100 to 101, the increase is much smaller.
Therefore, the model captures a situation where increasing x continues to increase y, but each additional increase in x produces a progressively smaller change in y.
Logarithmic Regression vs Linear Regression
A linear regression model has the form:
y = a + bx
The relationship between x and y is represented by a straight line.
A logarithmic regression model has the form:
y = a + b ln(x)
The relationship is curved.
Linear regression assumes a constant change in y for each one-unit increase in x, whereas logarithmic regression represents a relationship in which the effect of x becomes smaller as x increases.
Logarithmic Regression vs Polynomial Regression
Polynomial regression can model many types of curved relationships, while logarithmic regression is more specifically suited to patterns that change rapidly and then flatten.
For example:
Polynomial:
y = a + bx + cx²
Logarithmic:
y = a + b ln(x)
The best choice depends on the observed pattern and the purpose of the analysis.
Common Mistakes When Using Logarithmic Regression
1. Using zero or negative x-values
Because the model uses ln(x), x must be positive.
2. Choosing the model only because it has a high R²
A high R² does not automatically mean the model is appropriate.
3. Ignoring the scatterplot
Always inspect the data before fitting a regression model.
4. Extrapolating beyond the observed data
Predictions far outside the range of the original dataset can be unreliable.
5. Confusing logarithmic and exponential regression
Remember:
Logarithmic: y = a + b ln(x)
Exponential: y = abˣ
They describe different types of nonlinear relationships.
How to Find the Predicted Value on a TI-84
After obtaining the logarithmic regression equation, you can use the equation directly to make predictions.
For example, if the model is:
y = 76.21296 − 29.8634ln(x)
and you want to calculate the predicted value at x = 10, enter:
76.21296 − 29.8634ln(10)
The TI-84 will return the predicted y-value.
Alternatively, if the regression equation has been stored in Y1, you can enter the desired x-value in the calculator’s Y= or graphing environment to evaluate the fitted model.
Online Logarithmic Regression Calculator
If you do not have access to a TI-84 calculator, an online logarithmic regression calculator can be used to calculate the regression equation from a dataset.
An online calculator can be useful for:
- Finding the regression equation
- Estimating predicted values
- Comparing fitted models
- Working with larger datasets
- Checking calculations performed on a TI-84
When using a calculator, always verify that the x-values are valid for a logarithmic transformation.
Frequently Asked Questions
What is logarithmic regression?
Logarithmic regression is a nonlinear regression method that models the relationship between x and y using an equation such as y = a + b ln(x).
How do you perform logarithmic regression on a TI-84?
Enter your x-values in L1 and y-values in L2. Then select STAT → CALC → LnReg and press ENTER to calculate the logarithmic regression model.
What does LnReg mean on a TI-84?
LnReg is the TI-84 calculator’s logarithmic regression function. It fits a model of the form y = a + b ln(x) to the entered data.
What is the difference between logarithmic and exponential regression?
Logarithmic regression has the form y = a + b ln(x), while exponential regression has the form y = abˣ. Logarithmic regression generally represents a relationship that changes rapidly and then levels off, while exponential regression represents growth or decay based on a multiplicative factor.
Can logarithmic regression be used for negative x-values?
No. The standard real-valued logarithmic regression model requires x > 0 because ln(x) is undefined for zero and negative values.
Can logarithmic regression model both increasing and decreasing relationships?
Yes. If b > 0, the fitted relationship increases as x increases. If b < 0, the fitted relationship decreases as x increases.
Is logarithmic regression linear?
The relationship between x and y is nonlinear because it contains ln(x). However, it is linear in the transformed predictor ln(x).
Conclusion
Logarithmic regression is useful for analyzing nonlinear relationships where the response changes rapidly initially and then gradually levels off. The standard model is:
y = a + b ln(x)
On a TI-84 calculator, you can perform logarithmic regression by entering the data into L1 and L2, selecting STAT → CALC → LnReg, and examining the resulting coefficients.
The model can then be used to make predictions, graph the fitted relationship, and understand whether the relationship is increasing or decreasing. Because logarithmic regression requires positive x-values, always check your data before performing the analysis.
For more statistical calculator tutorials, explore the TI-84 Archives » FINNSTATS
