2.14 Logarithmic Function Context and Data Modeling

TLDR

Logarithmic functions model situations with proportional growth or repeated multiplication, where input values change proportionally while output values change in equal-length steps. In AP Precalculus, you build a logarithmic model from two input-output pairs, from a proportion and a real zero, or by using logarithmic regression with technology, then use that model to predict values for the dependent variable.

Why This Matters for the AP Precalculus Exam

Topic 2.14 connects logarithmic functions to real data and real contexts, which is exactly the kind of modeling thinking the AP Precalculus exam rewards. Because logarithmic functions are the inverses of exponential functions, this topic asks you to flip your perspective: instead of output values changing proportionally, the input values change proportionally over equal-length output-value intervals.

On the exam, you may need to:

• Recognize when a logarithmic model fits a context or data set.
• Build a model in the form $f(x) = a \log_b x$ using given information.
• Use logarithmic regression on a calculator and interpret the result.
• Predict values for the dependent variable and explain what your answer means in context.

Some questions allow a graphing calculator, so being fluent with entering data, running a logarithmic regression, and reading the output is worth practicing before test day. Showing clear steps is important for clear exam work, especially since small errors slip in easily when you manipulate logarithms.

Key Takeaways

• Logarithmic functions model proportional growth or repeated multiplication, with input values changing proportionally over equal-length output-value intervals.
• A logarithmic model can be built from a proportion and a real zero, or from two input-output pairs.
• You construct models by applying transformations to the general form $f(x) = a \log_b x$ based on the context or data.
• Technology lets you build a model from data using logarithmic regression.
• The natural logarithm $\ln x$ is often useful for modeling real-world phenomena.
• Once you have a model, you can predict values for the dependent variable.

Understanding Logarithmic Models

A logarithmic function tells you the exponent: $\log_b x$ is the power you raise the base $b$ to in order to get $x$. Because logarithmic functions are inverses of exponential functions, they fit situations where the output grows by equal amounts while the input is multiplied by a constant factor.

In other words, exponential growth has output values changing multiplicatively as input changes additively. Logarithmic growth flips this: output values change additively as input values change multiplicatively. That is why logarithmic models show up for things like sound level, earthquake magnitude, and acidity, where the underlying quantities span a huge range but the scale moves in steady steps.

If the output value happens to be a whole number, it tells you how many times the initial value has been multiplied by the proportion.

Sample Application: The Sound of Music

The decibel scale is a real-world application of a logarithmic model, not required AP content, but it shows how these functions work.

Sound level is often measured in decibels (dB), which are logarithmically related to the intensity of a sound. The intensity of a sound is measured in watts per square meter ($W/m^2$), and the relationship between intensity and decibels is given by the equation:

$L = 10 \log_{10} (I/I_0)$

where L is the sound level in decibels, I is the intensity of the sound, and $I_0$ is a reference intensity, typically taken to be the threshold of human hearing, which is approximately $10^{-12}\ W/m^2$.

Suppose we want to model the sound level of a rock concert in a large stadium. We measure the sound intensity at various locations and record the following data:

To find the sound level in decibels at each location, plug the measured intensity and reference intensity into the equation:

$L = 10 \log_{10} (I/I_0)$

For the stage: $L = 10 \log_{10} (1.0/10^{-12}) = 120\ dB$

For the front row: $L = 10 \log_{10} (0.1/10^{-12}) = 110\ dB$

For the middle of the crowd: $L = 10 \log_{10} (0.01/10^{-12}) = 100\ dB$

For the back of the crowd: $L = 10 \log_{10} (0.001/10^{-12}) = 90\ dB$

The sound level decreases as we move away from the stage, with a difference of 10 dB between each location. That makes sense, since sound intensity drops as distance from the source increases.

Now we can build a logarithmic function to model this relationship using the form:

$y = m \cdot \log_b(x) + c$

where y is the sound level, $x$ is the intensity, and $m$ and $c$ are constants.

From the data we have the points $(x, y)$:

$(1, 120),\ (0.1, 110),\ (0.01, 100),\ (0.001, 90)$

Using these points to solve for $m$ and $c$:

$m = 10$

$c = 0$

So the logarithmic function model is:

$y = 10 \cdot \log_{10}(x)$

We can use this function to predict the sound level at any intensity. For example, at $0.0001\ W/m^2$:

$y = 10 \cdot \log_{10}(0.0001) = 80\ dB$

This model gives us a way to quantify the relationship between sound intensity and sound level in decibels.

Lines and Logarithmic Regression

You can build a logarithmic model from an appropriate proportion and a real zero, or from two input-output pairs. With two points, set up a system using the form $f(x) = a \log_b x$ (or $y = m \log_b x + c$) and solve for the constants.

Technology, such as logarithmic regression, can also construct a logarithmic model for a full data set. The natural log function, $y = \ln(x)$, is often useful for modeling real-world phenomena like growth and decay.

Example of logarithmic regression with two graphs plotted: $y = 21.275\ln(x) - 20.255$ and $R^2 = 0.9768$. Image Courtesy of Micro PedSim

Once you have a model, you can predict values for the dependent variable based on the context or data set. To check that a model fits well, look at the residuals: a residual plot without a clear pattern supports your choice of model.

How to Use This on the AP Precalculus Exam

Problem Solving

• To build a model from two input-output pairs, plug both points into $f(x) = a \log_b x$ (or $y = m \log_b x + c$) and solve the resulting system for the constants.
• Watch the domain. A logarithmic function in general form is only defined for $x > 0$, and its graph has a vertical asymptote at $x = 0$.
• Track whether your model is increasing or decreasing, and whether it is concave up or concave down, since logarithmic functions are always one or the other.

Calculator

• Enter the data, run a logarithmic regression, and read off the constants for $y = a + b \ln x$ form.
• Use the regression equation to predict a dependent value, then state what that prediction means in the context.
• If you are deciding between models, plot the residuals and choose the model whose residual plot has no clear pattern.

Common Trap

• Do not assume every increasing data set is logarithmic. Logarithmic growth rises quickly at first and then levels off, showing diminishing returns, which is different from exponential growth.
• Keep your steps visible. Logarithm rules are easy to misapply, so write each manipulation out instead of skipping ahead.

Common Misconceptions

• Logarithmic and exponential are not the same thing. They are inverses. Exponential output changes multiplicatively as input changes additively; logarithmic output changes additively as input changes multiplicatively.
• A logarithmic model is not defined for all x. The general form $f(x) = a \log_b x$ requires $x > 0$, so predictions at or below zero do not make sense.
• A high $R^2$ alone does not prove the best model. You still check the residual plot for a pattern before deciding a logarithmic model is appropriate.
• The base matters, but all log functions are related. Because of the change of base property, every logarithmic function is a vertical dilation of every other, so you can convert between bases when needed.
• "Whole number output" has meaning. When the output of a logarithmic model is a whole number, it tells you how many times the initial value has been multiplied by the proportion, not a random count.

Related AP Precalculus Guides

• 2.8 Inverse Functions
• 2.10 Inverses of Exponential Functions
• 2.4 Exponential Function Manipulation
• 2.7 Composition of Functions
• 2.11 Logarithmic Functions
• 2.3 Exponential Functions