Unit 4 · Topic 2.14

Logarithmic Function Context and Data Modeling

Use logarithmic models and inverse operations to answer when or how many scale questions. Develop the idea through symbolic, numerical, graphical, and contextual representations.

Learning Goals

Recognize additive output change under multiplicative input change.
Construct logarithmic models from points, transformations, or regression.
Apply and validate models in context.

1. Essential Structure

Use logarithmic models and inverse operations to answer when or how many scale questions.

\[t=\frac{\ln\left(\frac{A}{A_0}\right)}{\ln b}\]

Read the formula together with its domain, units, starting input, and the interval length over which change is measured.

2. Core Ideas

A logarithmic model is supported when multiplying input by a fixed factor adds a fixed output amount.
In \(a\log_b(x/h)+k\), \(h\) is a reference input and \(k\) is its output.
Two points can determine parameters once a base or structural condition is chosen.
All contextual inputs must remain positive after any horizontal shift.

3. Graph and Representation

A logarithmic inverse converts multiplicative output levels back into elapsed time.

Use graphical evidence together with an algebraic or numerical reason. A graphing window alone does not establish domain, end behavior, or model validity.

4. Original Worked Example

For \(A(t)=100(2)^t\), when does \(A=800\)?

\(\frac{800}{100}=8=2^3\), so \(t=3\).

Check the result in the original representation and explain its meaning, including units when the quantities are contextual.

5. AP Reasoning Workflow

Define positive input and output quantities.
Look for proportional input and additive output patterns.
Choose a reference point and base.
Determine scale and shift.
Validate, apply, and interpret.

A strong AP response shows the mathematical evidence first and then states a precise conclusion.

6. Extended Example and Application

The model \(L(I)=10\log_{10}(I/I_0)\) adds 10 output units whenever intensity is multiplied by 10 and adds 20 when intensity is multiplied by 100.

7. Technology and Validation

Use logarithmic regression for noisy data and inspect residuals before extrapolating.

When technology is used, record the model or result, preserve sufficient precision, and explain why the output answers the question.

8. Recognize a Logarithmic Pattern in Data

A logarithmic model is appropriate when multiplying the input by a fixed factor changes the output by an approximately fixed amount. Compare output differences over proportional input intervals.

Input \(x\)	Output \(y\)	Input ratio	Output difference
2	4	--	--
6	7	3	3
18	10	3	3
54	13	3	3

Each multiplication of the input by 3 adds 3 to the output. An anchored model is

\[y=4+3\log_3\left(\frac{x}{2}\right).\]

At \(x=2\), the logarithm is zero and the output is 4. Replacing \(x\) by \(3x\) increases the logarithm by 1, so the model output rises by 3.

9. Build a Model from a Real Zero and a Proportion

Suppose a logarithmic quantity equals zero at input \(x=5\), and multiplying the input by 4 increases the output by 6. Use the real zero as the reference input:

\[L(x)=6\log_4\left(\frac{x}{5}\right).\]

The model satisfies

\[L(5)=0,\qquad L(20)=6,\qquad L(80)=12.\]

More generally, if \((h,k)\) is a known point and multiplication of the input by \(r\) changes the output by \(d\), then one useful form is

\[f(x)=k+d\log_r\left(\frac{x}{h}\right),\qquad x>0,\ h>0.\]

This form displays the reference pair and multiplicative input step directly.

10. Construct a Model from Two Input-Output Pairs

Suppose a logarithmic response passes through \((2,5)\) and \((8,11)\). The input changes by a factor of \(8/2=4\), while the output changes by \(11-5=6\). Anchoring at the first point gives

\[S(x)=5+6\log_4\left(\frac{x}{2}\right).\]

Check the second point:

\[S(8)=5+6\log_4(4)=11.\]

The same function may be written with natural logarithms:

\[S(x)=5+\frac{6}{\ln4}\ln\left(\frac{x}{2}\right).\]

Changing the displayed logarithm base changes the coefficient but not the modeled function. Two data points determine the slope with respect to \(\ln x\); the anchored form avoids treating a coefficient as a universal contextual parameter.

11. Model Data with a Horizontal Shift

Sometimes the raw inputs are not proportional, but their distances from a boundary are. Consider:

Input \(x\)	Output \(y\)	Adjusted input \(x-5\)
6	0	1
7	2	2
9	4	4
13	6	8

The adjusted inputs double whenever the output rises by 2, so

\[y=2\log_2(x-5),\qquad x>5.\]

The value 5 is the input boundary and vertical asymptote. It is not a permitted input. A horizontal shift should be supported by a meaningful threshold, baseline, or pattern in the adjusted data rather than introduced only to improve visual fit.

12. Use Logarithmic Regression with Noisy Data

Observed data rarely follow an exact logarithmic pattern. If a scatterplot rises or falls quickly and then levels gradually, technology can fit a model of the form \(y=A+B\ln x\), with \(x>0\).

\(x\)	1	2	4	8	16
Observed \(y\)	12.5	16.3	20.2	24.4	28.5

Logarithmic regression gives approximately

\[\hat y=12.36+5.785\ln x.\]

The residuals for the listed points are approximately \(0.14,-0.07,-0.18,0.01,0.10\), a small patternless set around zero. Keep the unrounded regression coefficients for calculations, then round the final prediction.

Plot the original data and check the positive input domain.
Fit a logarithmic regression.
Inspect residual magnitude and pattern.
Interpret predictions only over a contextually reasonable interval.

13. Use the Natural Logarithm in Models

The natural logarithm is often convenient in scientific and data models:

\[f(x)=A+B\ln x,\qquad x>0.\]

It is not a different model family from other logarithm bases. By change of base,

\[A+B\ln x=A+(B\ln b)\log_bx.\]

Thus changing the base changes the numerical coefficient while preserving the same curve.

For the regression model \(\hat y=12.36+5.785\ln x\), multiplying the input by 2 changes the predicted output by

\[5.785\ln(2x)-5.785\ln x=5.785\ln2\approx4.01.\]

This is a model-specific change statement. The AP course does not require a universal contextual interpretation of the coefficients in every representation; use the stated variables, units, and input factor.

14. Predict Outputs and Determine Inputs

A logarithmic model can answer both forward and inverse questions. Suppose a response score is modeled by

\[S(x)=5+6\log_4\left(\frac{x}{2}\right),\qquad 2\le x\le32.\]

Predict an output:

\[S(32)=5+6\log_4(16)=17.\]

Determine an input: when \(S(x)=14\),

\[14=5+6\log_4(x/2)\]

\[\log_4(x/2)=\frac32\Longrightarrow\frac{x}{2}=4^{3/2}=8\Longrightarrow x=16.\]

The model predicts a score of 17 at input 32 and reaches score 14 at input 16.

Reporting check	Question
Mathematical domain	Is the adjusted logarithm argument positive?
Contextual domain	Does the input lie in the measured or meaningful interval?
Units	Are both the predicted quantity and input labeled?
Extrapolation	Could the leveling pattern change beyond the data?

15. Common Errors

Interpret the logarithm's output in the units of the original input variable.
Using a logarithmic model for constant additive input change.
Ignoring positivity restrictions.
Giving a numerical result without a domain check, units, or interpretation.

Key Takeaways

Use logarithmic models and inverse operations to answer when or how many scale questions.
Core relationship: \(t=\frac{\ln\left(\frac{A}{A_0}\right)}{\ln b}\)
Error check: Interpret the logarithm's output in the units of the original input variable.

Checkpoint · Topic 2.14

Construct \(a\log_2x+k\) through \((1,3)\) and \((8,15)\).
Interpret doubling the input in your model.
Explain how residuals compare logarithmic and linear regressions.