Unit 3 · Topic 2.5
Exponential Function Context and Data Modeling
Construct exponential models from initial values, ratios, percent changes, or regression output. Develop the idea through symbolic, numerical, graphical, and contextual representations.
Learning Goals
- Construct exponential models from factors, rates, two points, and regression.
- Use doubling time and half-life forms.
- Interpret predictions with units, domains, and limitations.
1. Essential Structure
Construct exponential models from initial values, ratios, percent changes, or regression output.
Read the formula together with its domain, units, starting input, and the interval length over which change is measured.
2. Core Ideas
- A nearly constant output ratio over equal input steps supports an exponential model.
- From two positive-output points, \(b=(y_2/y_1)^{1/(x_2-x_1)}\).
- Growth rate \(r\) gives factor \(1+r\); decay rate \(r\) gives \(1-r\).
- Doubling time \(D\) gives \(P(t)=P_0 2^{t/D}\), while half-life \(H\) gives \(P(t)=P_0(1/2)^{t/H}\).
3. Graph and Representation
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
Interpret \(P(t)=200(1.04)^t\).
Answer: The initial value is 200 and the quantity grows 4% per time unit.
Check the result in the original representation and explain its meaning, including units when the quantities are contextual.
5. AP Reasoning Workflow
- Define variables and units.
- Verify proportional change.
- Choose an anchor and determine the factor.
- Write and check the model.
- Apply it and interpret within the contextual domain.
A strong AP response shows the mathematical evidence first and then states a precise conclusion.
6. Extended Example and Application
A medication amount is 180 mg at hour 2 and 90 mg at hour 8. A model is \(M(t)=180(1/2)^{(t-2)/6}\), which predicts 45 mg at hour 14.
7. Technology and Validation
For imperfect data, use exponential regression and inspect residuals. Retain coefficient precision until the final answer.
When technology is used, record the model or result, preserve sufficient precision, and explain why the output answers the question.
8. Decide Whether Exponential Modeling Is Appropriate
An exponential model is supported when outputs are multiplied by approximately the same factor over equal-length input intervals. Compare ratios, not raw differences.
| Year \(t\) | Accounts \(A\) | Difference | Successive ratio |
|---|---|---|---|
| 0 | 500 | -- | -- |
| 1 | 575 | 75 | 1.15 |
| 2 | 661.25 | 86.25 | 1.15 |
| 3 | 760.44 | 99.19 | about 1.15 |
The differences grow, but the ratios remain nearly constant, so \(A(t)=500(1.15)^t\) is appropriate. A ratio of 1 would describe a constant output, and a ratio of 0 would not define a genuine exponential pattern.
9. Build a Model from an Initial Value and a Rate
For \(Q(t)=Q_0b^t\), \(Q_0\) is the value at \(t=0\) and \(b\) is the one-unit factor. Convert a percent change into a factor before writing the model.
| Change per unit | Factor | Model |
|---|---|---|
| Increase by \(r\) | \(1+r\) | \(Q_0(1+r)^t\) |
| Decrease by \(r\) | \(1-r\) | \(Q_0(1-r)^t\) |
| Multiply by \(R\) every \(h\) units | \(R^{1/h}\) per unit | \(Q_0R^{t/h}\) |
Example: A device worth $1,800 loses 18% of its value each year. It retains \(1-0.18=0.82\), so
After 4 years, \(V(4)\approx813.82\). The model predicts a value of about $814, subject to the assumptions of constant proportional depreciation and the practical lifetime of the device.
10. Construct an Exponential Model from Two Points
Suppose \(f(x)=ab^x\) passes through \((x_1,y_1)\) and \((x_2,y_2)\), where the relevant outputs are positive. Dividing the two equations eliminates \(a\):
Worked example: A colony contains 320 cells at hour 2 and 1,080 cells at hour 5. Then
Using the first point as an anchor gives
An equivalent general form is \(C(t)=\frac{1280}{9}(1.5)^t\). Substitution of \(t=2\) and \(t=5\) confirms both given values. The anchored form is usually clearer because its coefficient has a direct contextual meaning.
11. Reveal a Shifted Exponential Pattern
Some data approach a nonzero baseline \(k\). The original outputs may not have a constant ratio, but the adjusted outputs \(y-k\) can. This leads to a transformed model
| Time \(t\) | Temperature \(T\) | Excess \(T-20\) | Excess ratio |
|---|---|---|---|
| 0 | 100 | 80 | -- |
| 1 | 68 | 48 | 0.60 |
| 2 | 48.8 | 28.8 | 0.60 |
| 3 | 37.28 | 17.28 | 0.60 |
The temperature itself does not have a constant ratio. Its excess above the 20-degree surroundings does, giving
The model approaches 20 rather than 0. In context, the added constant represents the ambient level and the horizontal asymptote.
12. Connect Percent Factors, the Natural Base, and Time Scales
The same model can use different bases. If \(Q(t)=Q_0b^t\), then \(b=e^k\) with \(k=\ln b\), so
For \(Q(t)=750(1.06)^t\), the continuous-rate form is \(Q(t)=750e^{(\ln1.06)t}\approx750e^{0.05827t}\). The value 0.05827 is not the same as a 5.827% discrete increase; it is the continuous growth constant that produces the same one-unit factor.
Equivalent forms can also reveal different time intervals:
If \(t\) is measured in months, the last form shows that the quantity is multiplied by about 2.0122 every 12 months.
13. Use Exponential Regression with Data
Real data rarely have perfectly constant ratios. When the scatterplot and context suggest proportional change, technology can fit \(y=ab^x\) by exponential regression.
- Enter paired input-output data and make a scatterplot.
- Check that an exponential model is plausible in the contextual domain.
- Run exponential regression and record \(a\) and \(b\) with sufficient precision.
- Interpret \(a\) only if \(x=0\) is meaningful; interpret \(b\) as the estimated one-unit factor.
- Use the unrounded regression coefficients for predictions, then round the final contextual answer.
Suppose technology returns \(P(t)=246.8(1.0736)^t\) for a town's population in thousands, where \(t\) is years since 2020. The model estimates 246.8 thousand residents in 2020 and annual growth of \((1.0736-1)100\%=7.36%\).
14. Apply the Model and Respect Its Domain
An exponential model can answer a forward question, such as finding an output, or an inverse question, such as finding when a threshold is reached. Always attach units and consider whether the input is allowed by the context.
Forward prediction: For \(B(t)=1200(1.025)^t\), where \(t\) is years after 2025,
The model predicts about 1,462 items in 2033.
Threshold question: To determine when the amount first exceeds 1,500, solve
A graph or numerical solver gives \(t>9.04\). If measurements occur only at the end of each whole year, the first reported year is \(t=10\), or 2035.
| Check before reporting | Question to ask |
|---|---|
| Units | Does the exponent use the same time unit as the growth factor? |
| Domain | Are fractional or negative inputs meaningful? |
| Rounding | Must the answer be a whole person, payment period, or measurement time? |
| Extrapolation | Is constant proportional change still reasonable beyond the observed data? |
15. Common Errors
- Do not use 0.04 as the exponential base; use the growth factor 1.04.
- Using the percent rate itself as the exponential base.
- Extrapolating without discussing whether the model remains plausible.
- Giving a numerical result without a domain check, units, or interpretation.
Key Takeaways
- Construct exponential models from initial values, ratios, percent changes, or regression output.
- Core relationship: \(P(t)=P_0(1+r)^t\)
- Error check: Do not use 0.04 as the exponential base; use the growth factor 1.04.
- Model a population of 1,200 growing 2.5% yearly.
- A value changes from 40 to 135 over six units. Find its one-unit factor.
- Explain why constant half-life is evidence for exponential decay.