Algebra 3 Quiz: Fitting Models To Data

11 questions · exam conditions

0:00

Fitting Models To DataQuestion 1 of 11

A biologist fits the model $N(t)=500(1.32)^{t/4}$ to population counts, where $t$ is measured in hours. Which of the following best gives the hourly percent increase in the population?

About 7.2% per hour

About 8.0% per hour

About 32% per hour

About 0.32% per hour

← Back to quizzes

Algebra 3 Quiz

Algebra 3 Quiz: Fitting Models To Data

Practice Fitting Models To Data in Algebra 3 with focused quiz questions that help you check what you know, review explanations, and build confidence with test-style prompts.

What this quiz covers

This quiz focuses on Fitting Models To Data, giving you a quick way to practice the rules, question types, and explanations that matter most for Algebra 3.

How to use this quiz

Try each quiz question before looking at the correct answer. Use the explanations to review missed ideas, then come back to similar questions until the pattern feels familiar.

All questions

Question 1

A biologist fits the model $N(t)=500(1.32)^{t/4}$ to population counts, where $t$ is measured in hours. Which of the following best gives the hourly percent increase in the population?

About 7.2% per hour (correct answer)
About 8.0% per hour
About 32% per hour
About 0.32% per hour

Explanation: When you see a growth model like

N(t)=500(1.32)^{t/4}

, the base

1.32

is the growth factor for the time interval shown in the exponent's denominator:

t/4

means the multiplier applies once every 4 hours. To find the hourly percent increase, convert the 4-hour growth factor to a 1-hour factor: hourly factor

=1.32^{1/4}

. Using a calculator,

1.32^{1/4}\approx 1.0717

, so each hour the population is multiplied by about

1.0717

. An increase from

1

1.0717

is a

7.2\%

gain, so the correct answer is About 7.2% per hour. Why not the others? About 32% per hour treats the 32% increase as if it happened every hour, but the

t/4

exponent means it happens over 4 hours. About 8.0% per hour comes from dividing 32% by 4, but exponential growth is not linear; you need the fourth root, not division. About 0.32% per hour likely confuses the decimal

0.32

with

0.32\%

, but

0.32

1.32

represents 32 percentage points, not

0.32\%

. On exam day, when the growth interval is longer than one unit, raise the base to

1/(\text{number of units})

to get the per-unit factor. If the model were

(1.32)^t

, the hourly increase would be 32%—but the

/4

changes everything.

Question 2

An auto-safety researcher fits the quadratic model $d=0.06s^2+0.7s+5$ to data on stopping distance $d$ in feet for a car traveling at speed $s$ in mph, for speeds between 10 and 60 mph. According to the model, what is the predicted increase in stopping distance when speed increases from 30 mph to 40 mph?

10 ft
42 ft
49 ft (correct answer)
129 ft

Explanation: When you see a question asking for a "predicted increase" from a model, focus on the word increase: you need the difference between two outputs, not just one output. Here, that means evaluating the quadratic model at both speeds and subtracting. At 30 mph:

d=0.06(30)^2+0.7(30)+5=54+21+5=80\text{ ft}

At 40 mph:

d=0.06(40)^2+0.7(40)+5=96+28+5=129\text{ ft}

The predicted increase is therefore:

129-80=49\text{ ft}

So the choice "49 ft" is correct. Now look at the traps. "10 ft" is simply the increase in speed itself, 40 minus 30, but the model is nonlinear, so distance does not increase by one number per mph. "42 ft" comes from computing only the quadratic part of the change:

0.06(40^2-30^2)=42

, which ignores the linear term and the constant term. And "129 ft" is the total stopping distance at 40 mph, not the increase from 30 mph. A strong strategy: whenever a question says "increase," write down the two quantities you need to subtract before doing the arithmetic. For quadratic models especially, don't confuse a single output value with a change in output, and don't assume the change is proportional to the speed change. Evaluating the model at each endpoint and subtracting will always keep you on track.

Question 3

A student fits a least-squares linear model to a data set. The residuals show a clear curved pattern, negative at low and high x-values and positive in the middle. Which next step is most appropriate?

Conclude that the linear model is appropriate because the residuals are small.
Add a quadratic term or try a quadratic model. (correct answer)
Remove the point with the largest residual and refit the linear model.
Take the reciprocal of the response variable and refit a linear model.

Explanation: When you see a residual plot with a clear shape, your first thought should be: the pattern is information. Residuals are the leftovers after the linear model — if they show a systematic curve, negative at low and high

x

-values and positive in the middle, that's a signature of a curved relationship. A linear model cannot capture that bend, so the most appropriate next step is to add a quadratic term or try a quadratic model. This directly models the curvature, allowing the fitted values to follow the data instead of missing the middle and ends. Why not the other choices? "Conclude that the linear model is appropriate because the residuals are small" ignores the fact that a clear curved pattern is not random noise; even small residuals with structure mean the model is systematically wrong. "Remove the point with the largest residual and refit" treats the issue as an outlier, but the pattern is spread across the whole data set, not caused by one unusual point — removing one point won't fix the bend. "Take the reciprocal of the response variable" is a transformation that can help with certain nonlinear forms, but the symmetric U-shaped residual pattern is much more directly addressed by adding a quadratic term than by reciprocating the response. Your study tip: when reading a residual plot, always ask whether the residuals look "boring" — random, no shape. If you see a curve, a funnel, or a trend, the model is missing something. A curve in residuals = time to add curvature.

Question 4

A city transportation office fits the linear regression $y=25x+120$ to predict daily bicycle rentals $y$ from high temperature $x$ in degrees Fahrenheit, using data for days with $x$ between 50°F and 90°F. Which interpretation of the intercept 120 is most appropriate?

On a 0°F day, the model predicts 120 rentals.
The intercept is not practically meaningful because 0°F is outside the data range. (correct answer)
There are 120 rentals on an average day regardless of temperature.
For each degree increase, predicted rentals increase by 120.

Explanation: Whenever you see a linear regression equation, ask yourself what each coefficient actually represents: the slope tells you the predicted change in

y

per one-unit change in

x

, and the intercept is the predicted

y

when

x=0

. Here, the model is

\hat y = 25x + 120

, but the data only cover temperatures from 50°F to 90°F. That is why the best interpretation of 120 is that the intercept is not practically meaningful because 0°F is outside the data range. Plugging

x=0

into the equation would give 120 rentals, but doing so extrapolates far beyond the temperatures used to build the model, so that number has no real-world credibility. The choice saying "On a 0°F day, the model predicts 120 rentals" is a trap: the arithmetic is correct, but it ignores the crucial limitation that 0°F is not in the observed range. "There are 120 rentals on an average day regardless of temperature" confuses the intercept with the overall average of

y

; the model does not claim rentals are constant. Finally, "For each degree increase, predicted rentals increase by 120" misidentifies the slope—the actual slope is 25, not 120. A strong study habit: when interpreting an intercept, always check whether

x=0

falls inside the data range. If it does not, say the intercept is not practically meaningful. Also, remember slope is the rate of change, while intercept is the starting value at

x=0

Question 5

A radioactive sample initially contains 80 mg. After 6 hours, 20 mg remains. Assuming the decay is exponential, what is the half-life of the substance?

1.5 hours
12 hours
6 hours
3 hours (correct answer)

Explanation: Whenever you see exponential decay, remember the half-life is the time for the amount to be cut in half. Here the sample drops from 80 mg to 20 mg in 6 hours. Notice that 20 mg is one-fourth of 80 mg, and since each half-life multiplies the amount by

\frac{1}{2}

, two half-lives have passed:

\frac{1}{2}\times\frac{1}{2}=\frac{1}{4}

. So those 6 hours represent 2 half-lives, meaning each half-life is

\frac{6}{2}=3

hours. The half-life is 3 hours. The 6 hours choice is a trap: the decay time given is not itself the half-life because the sample didn't merely halve from 80 to 40; it went all the way to 20. The 12 hours choice comes from incorrectly thinking that because

\frac{1}{4}

is two halves, you should multiply

6 \times 2

to get 12. But that would mean the half-life is longer than the total time observed, which is impossible here since the sample already fell to one-fourth in 6 hours. The 1.5 hours choice divides 6 by 4, confusing "one-fourth of the original amount" with "four half-lives." A quarter remaining is exactly two half-lives, not four. A reliable strategy: write

A = A_0\left(\frac{1}{2}\right)^{t/h}

, plug in the values, and solve for

h

. Recognizing that

80 \to 20

is a factor of

\frac{1}{4}

is the fastest path:

\frac{1}{4} = \left(\frac{1}{2}\right)^2

, so 6 hours equals two half-lives. Always ask yourself: "How many half-lives fit into the given time?"

Question 6

A scientist believes two quantities x and y are related by a power model $y=ax^b$ . After computing $L=\ln y$ and $M=\ln x$ , technology gives the least-squares line $L=2.1+0.7M$ . Which model best fits the original data?

$y = e^{2.1} + x^{0.7}$
$y = 0.7x + 2.1$
$y = 2.1e^{0.7x}$
$y = e^{2.1}x^{0.7}$ (correct answer)

Explanation: When you see a least-squares line after taking logarithms of both variables, you are dealing with a log-log transformation. That is the key to recovering a power model. Starting from

y=ax^b

, taking the natural log gives

\ln y = \ln a + b\ln x

. This is a linear equation in the transformed variables, where the intercept is

\ln a

and the slope is

b

. Here the line is

L=2.1+0.7M

, so the intercept

2.1=\ln a

and the slope

0.7=b

. Exponentiating the intercept gives

a=e^{2.1}

, while the power stays

0.7

. Therefore the original model is

y=e^{2.1}x^{0.7}

. Each wrong answer comes from a common misreading. The choice

y=e^{2.1}+x^{0.7}

treats the log-linear pieces as if they were added in the original scale — but the model is multiplicative, not additive. The choice

y=0.7x+2.1

incorrectly treats the coefficients as if they were already the original linear model, ignoring the logarithms entirely. Finally,

y=2.1e^{0.7x}

mistakes the log-log relationship for an exponential one; that would correspond to a line like

\ln y = \ln 2.1 + 0.7x

, not a line in

\ln x

. Remember the pattern: when both variables are logged, exponentiate the intercept and keep the slope as the exponent. That instantly converts the line back to a power model.

Question 7

A least-squares linear regression using 30 data points gives y-hat = ̂? Wait, use: y-hat = ̂?

For each 1-unit increase in x, the predicted value of y increases by 3.2 units. (correct answer)
Because r = ̂? Wait,
The correlation is strong enough to conclude that x causes y.
The regression line must pass through the data point (0,15).

Explanation: Whenever you see a least-squares regression equation, ask yourself: what does the slope mean? In

\hat y = 15 + 3.2x

, the slope is

3.2

, so a 1-unit increase in

x

produces a predicted increase of

3.2

units in

y

. That is exactly the correct interpretation. The choice beginning "Because

r =

…" confuses the correlation coefficient with the slope.

r

measures strength and direction, not the rate of change in

y

per unit change in

x

. The statement about strong correlation proving causation is a classic trap: no matter how strong the correlation, it does not establish that

x

causes

y

. The choice about

(0,15)

confuses the

y

-intercept with an actual data point. The line crosses the

y

-axis at the predicted value when

x=0

, but the regression line is not required to pass through any observed data point; it always passes through

(\bar{x}, \bar{y})

. A useful habit: when interpreting regression output, always attach "predicted" to your language and focus on the slope as the change per one-unit increase in

x

. And remember: correlation is not causation.

Question 8

A data set is thought to follow an exponential relationship $y=ab^x$ . After computing $L=\ln y$ , a student uses technology to fit the least-squares line $L=1.02+0.41x$ . Which equation represents the original model?

$y = e^{1.02+0.41x}$ (correct answer)
$y = 1.02e^{0.41x}$
$y = e^{1.02} + e^{0.41x}$
$y = 1.02e^x + 0.41$

Explanation: Whenever you see a data set modeled as

y=ab^x

, think about log-linearization: taking the natural log of both sides gives

\ln y=\ln a+(\ln b)\,x

. So the least-squares line for

L=\ln y

has intercept

\ln a

and slope

\ln b

. Here the fitted line is

L=1.02+0.41x

. That means

\ln a=1.02

and

\ln b=0.41

. Exponentiating gives

a=e^{1.02}

and

b=e^{0.41}

. Substituting into

y=ab^x

yields

y=e^{1.02}(e^{0.41})^x=e^{1.02+0.41x}.

So the original model is the choice

y=e^{1.02+0.41x}

. The choice

y=1.02e^{0.41x}

treats

1.02

and

0.41

as if they were already

a

and

b

, but they are coefficients of the logged line, not the original parameters. The choice

e^{1.02}+e^{0.41x}

misapplies exponent rules:

e^{u+v}=e^u e^v

, not

e^u+e^v

. The choice

y=1.02e^x+0.41

treats the relationship as linear in

y

, ignoring that

L=\ln y

was what was fitted. Study tip: whenever you fit a line to

\ln y

, exponentiate the whole line to return to the original exponential model: if

\ln y=mx+b

, then

y=e^b(e^m)^x

Question 9

For $x=0,1,2,3,4$ , the corresponding y-values from an experiment are $2,5,10,17,26$ . Which type of model is best supported by the pattern in these data?

Linear, because the y-values are increasing.
Exponential, because the y-values grow faster as x increases.
Quadratic, because the first differences increase by a constant amount. (correct answer)
Power, because the ratios of consecutive y-values are nearly constant.

Explanation: When you're given a set of

(x,y)

values, don't just look at whether

y

is rising — look at the pattern of the differences. Here, the

y

-values are

2,5,10,17,26

. The first differences are

5-2=3

10-5=5

17-10=7

, and

26-17=9

. Those first differences are

3,5,7,9

, which increase by the constant amount

2

each time. A constant second difference is the signature of a quadratic model, so "Quadratic, because the first differences increase by a constant amount" is supported by the data. Now check the other choices. "Linear, because the y-values are increasing" confuses direction with form: a linear model requires constant first differences, not merely increasing values. "Exponential, because the y-values grow faster as x increases" is tempting because the values do grow faster, but exponential growth means the ratios of consecutive

y

-values are roughly constant. Here the ratios are

5/2=2.5

10/5=2

17/10=1.7

, and

26/17\approx1.53

, so they are not nearly constant. "Power, because the ratios of consecutive y-values are nearly constant" repeats that same ratio error; constant ratios point to exponential, not power, and these ratios aren't constant anyway. Study tip: when analyzing data, compute first differences. If they are constant, the model is linear; if the first differences themselves change by a constant amount, the model is quadratic. Save ratio checks for exponential questions.

Question 10

A psychologist fits the model $R=0.85-0.12\ln t$ to reaction time $R$ in seconds as a function of trial number $t$ , using data for $t \ge 10$ . Which statement is best supported by the model?

Each additional trial decreases predicted reaction time by 0.12 seconds.
As $t$ grows large, the predicted reaction time approaches 0 seconds.
When $t=1$ , the predicted reaction time is exactly 0 seconds.
Doubling the trial count decreases predicted reaction time by about 0.083 seconds. (correct answer)

Explanation: Whenever you see a model with a logarithm of time, the key is to think about multiplicative changes, not additive ones. The coefficient

-0.12

tells you how

R

changes when

\ln t

changes by 1, which corresponds to multiplying

t

e

, not adding one trial. . For the correct answer, compare the reaction time at trial

t

and at trial

2t

R(2t)-R(t) = 0.85-0.12\ln(2t) - (0.85-0.12\ln t) = -0.12\ln 2 \approx -0.083.

So doubling the trial count lowers predicted reaction time by about

0.083

seconds—exactly what the correct statement says. The other statements each misread the model. "Each additional trial decreases predicted reaction time by

0.12

seconds" confuses the coefficient on

\ln t

with a constant per-trial decrease; in a logarithmic model, the marginal effect of one extra trial shrinks as

t

grows. "As

t

grows large, the predicted reaction time approaches 0 seconds" is also wrong: since

\ln t

grows without bound,

R

keeps decreasing and will eventually become negative—it never settles at 0. "When

t=1

, the predicted reaction time is exactly 0 seconds" ignores the intercept:

R(1)=0.85-0.12\ln 1=0.85

, not 0. On this type of question, translate logarithmic changes into percentage or multiplicative changes: doubling, tripling, or multiplying by

e

are the natural increments to test. That habit will help you avoid the most common trap—treating logarithmic coefficients as if they applied to ordinary unit increases.

Question 11

A company fits the quadratic model $P=-0.4a^2+8a+20$ to annual profit $P$ in thousands of dollars as a function of advertising spending $a$ in thousands of dollars. According to the model, what level of advertising spending maximizes profit?

$8,000
$10,000 (correct answer)
$20,000
$60,000

Explanation: Whenever you see a quadratic model and a question about maximizing profit, think vertex: the negative coefficient on the squared term tells you the parabola opens downward, so the highest point on the curve is the vertex. Here, the model is

P=-0.4a^2+8a+20

, so use the vertex formula for the input value:

a=-\frac{8}{2(-0.4)}=-\frac{8}{-0.8}=10

. Since

a

is measured in thousands of dollars,

a=10

means

\

10{,}000$ in advertising spending. The other choices each come from misreading the model.

\

8{,}000

takes the linear coefficient

as if it were already the answer, but that is only part of the vertex calculation.

$20{,}000

is the constant term, which represents the profit when no advertising is spent, not the spending amount that maximizes profit.

$60{,}000

is the maximum profit itself: plug

a=10

into the model to get

P=-0.4(10)^2+8(10)+20=60

, meaning

$60{,}000$ in profit — but the question asks for advertising spending, not profit. So when you see a quadratic word problem, first identify what the question wants: the input that gives the maximum or the maximum output. For the input, compute

-\frac{b}{2a}

, then convert to the proper units. For a maximum profit, the answer should be the advertising level, not the profit value.