Building Function Models — I can build a function model...

SECTION 1

Historical Context & Motivation

Long before graphing calculators or spreadsheets existed, people needed to predict the future — how much grain to plant, how quickly a population would grow, or how far a cannonball would fly. Each of these questions demanded a different kind of mathematical relationship. Over centuries, mathematicians discovered that most real-world patterns fall into a surprisingly small number of function families, each with its own distinctive shape and behavior. Understanding which family fits a given situation is the foundation of mathematical modeling — the art of turning real life into equations we can analyze and solve.

~300 BCE

Euclid & Proportional Reasoning

Greek mathematicians formalized the idea of direct proportion — if you double the input, the output doubles. This is the earliest ancestor of linear functions.

1600s

Galileo & Quadratic Motion

Galileo showed that a falling object covers distance proportional to the square of elapsed time. This was one of the first real-world applications of quadratic functions in physics.

1798

Malthus & Exponential Growth

Thomas Malthus argued that human populations grow exponentially while food supply grows linearly, highlighting the explosive power of exponential functions.

1900s–Today

Modern Modeling

Scientists, economists, and engineers routinely choose among function families to model phenomena from viral spread to satellite trajectories. Choosing the right model is now a core skill in data science.

The central question this lesson addresses is: given a real-world context — a table of data, a verbal description, or a graph — how do you decide whether a linear, exponential, or quadratic model is the best fit, and how do you justify that choice? Answering this question turns you from someone who can solve equations into someone who can create them.

SECTION 2

Core Principles & Definitions

Before you can build a function model, you need to recognize the signatures of the three main function families. Each family has a unique algebraic form, a characteristic graph shape, and a telltale pattern in its data. When you encounter a new context, these signatures act like fingerprints that help you identify which family is at work.

Linear Functions

Form: y = mx + b. The graph is a straight line. The defining feature is a constant rate of change — equal changes in x always produce equal changes in y.

Exponential Functions

Form: y = a · bˣ. The graph curves sharply upward (growth) or toward zero (decay). The defining feature is a constant ratio (multiplicative change) — equal changes in x multiply y by the same factor.

Quadratic Functions

Form: y = ax² + bx + c. The graph is a parabola — a U-shape (or inverted U). The defining feature is that the rate of change itself changes at a constant rate — second differences are constant.

Function Model

A function model is an equation chosen to represent a real-world relationship. Building one means identifying input/output variables, selecting a function family, and writing the specific equation.

Justification

Simply writing an equation isn't enough. You must justify your choice using evidence — constant differences, constant ratios, graph shape, or the real-world behavior described in the context.

✦ KEY TAKEAWAY

Think of function families like vehicle types. A linear function is like a car on cruise control — steady speed, predictable distance. An exponential function is like a rocket — it keeps accelerating faster and faster (or slowing down by a fraction). A quadratic function is like a ball tossed in the air — it rises, peaks, and falls in a smooth arc. Picking the right model is like picking the right vehicle for the trip.

SECTION 3

Visual Explanation — The Three Function Families Side by Side

Side-by-side comparison of the three function families. Notice how the linear graph is a straight line, the exponential graph curves increasingly steeply, and the quadratic graph forms a symmetric U-shape (parabola) with a vertex.

The graph shapes above reveal each family's personality. A linear function makes perfectly even progress — no surprises. An exponential function starts slowly but then takes off dramatically (or, for decay, drops off sharply and levels out near zero). A quadratic function has a turning point: it either reaches a minimum (opens up) or a maximum (opens down), making it ideal for situations involving projectile motion, area optimization, or revenue modeling.

When you see data or read a description, ask yourself: does this scenario describe steady, unchanging growth? That's linear. Does the quantity multiply by the same factor each period? That's exponential. Does the quantity speed up, hit a peak or valley, and then reverse direction? That's quadratic. Matching the behavior of the context to these signatures is the heart of model-building.

SECTION 4

Mathematical Framework — Equations & Data Tests

Each function family has both a standard equation form and a diagnostic test you can apply to a table of data. These tests — based on differences and ratios — give you concrete numerical evidence for your choice of model.

LINEAR MODEL

y = mx + b

m = slope (constant rate of change), b = y-intercept (starting value when x = 0). Data test: first differences (Δy) are constant.

EXPONENTIAL MODEL

y = a · bˣ

a = initial value (y-intercept when x = 0), b = common ratio (growth factor if b > 1, decay factor if 0 < b < 1). Data test: consecutive ratios (y₂/y₁) are constant.

QUADRATIC MODEL

y = ax² + bx + c

a = leading coefficient (determines direction and width of parabola), b and c shift the parabola. Data test: second differences (Δ²y) are constant.

📐 What Are "Differences" and "Ratios"?

First differences are computed by subtracting consecutive y-values: Δy = y₂ − y₁. If every first difference is the same, the data is linear. Second differences are the differences of the first differences: Δ²y = Δy₂ − Δy₁. If these are constant, the data is quadratic. Consecutive ratios are computed by dividing consecutive y-values: y₂ ÷ y₁. If every ratio is the same, the data is exponential. These tests only work reliably when the x-values are equally spaced (e.g., x = 0, 1, 2, 3, …).

SECTION 5

Detailed Breakdown — Applying the Difference & Ratio Tests

Let's see the diagnostic tests in action with a concrete example. Suppose you're given three different data sets and you need to classify each one. The table below shows each data set along with its first differences, second differences, and consecutive ratios.

Each data set is analyzed using first differences, second differences, and consecutive ratios. The test that produces a constant result reveals the function family. Data Set A has constant first differences (linear), Data Set B has constant ratios (exponential), and Data Set C has constant second differences (quadratic).

Notice that the test that "wins" — the one that produces a constant — tells you the family. If none of the tests produce a perfectly constant result, the data may be approximately one type (choose the closest match) or may belong to a different family altogether (such as square root or logarithmic), which you'll study later.

💡 Context Clues Beyond Data

You don't always get a neat data table. Sometimes you only get a verbal description. Look for key phrases: "increases by the same amount each…" signals linear. "doubles every…" or "decreases by 15% each…" signals exponential. "maximum height," "area of a rectangle with fixed perimeter," or "launched into the air" signals quadratic.

SECTION 6

Worked Example — From Context to Model

A local coffee shop tracks the number of loyalty-app downloads each week after launching a social media campaign. Here is the data they collected:

Coffee shop loyalty-app download data

Week (x)	Downloads (y)
0	50
1	100
2	200
3	400
4	800

Building the Function Model

Step 1 — Identify VariablesThe independent variable (input) is x = weeks since the campaign launched. The dependent variable (output) is y = total downloads.

Step 2 — Compute First DifferencesΔy: 100 − 50 = 50, 200 − 100 = 100, 400 − 200 = 200, 800 − 400 = 400. The first differences are 50, 100, 200, 400 — NOT constant, so the data is not linear.

First differences: 50, 100, 200, 400 → Not constant → Not linear

Step 3 — Compute Second DifferencesΔ²y: 100 − 50 = 50, 200 − 100 = 100, 400 − 200 = 200. The second differences are 50, 100, 200 — NOT constant, so the data is not quadratic.

Second differences: 50, 100, 200 → Not constant → Not quadratic

Step 4 — Compute Consecutive Ratios100 ÷ 50 = 2, 200 ÷ 100 = 2, 400 ÷ 200 = 2, 800 ÷ 400 = 2. Every ratio equals 2 — CONSTANT. The data is exponential.

Consecutive ratios: 2, 2, 2, 2 → Constant → Exponential ✓

Step 5 — Write the EquationUsing y = a · bˣ, we identify a = 50 (the y-value when x = 0) and b = 2 (the common ratio). The model is:

y = 50 · 2ˣ

Step 6 — Justify the ChoiceWe chose an exponential model because the consecutive ratios are constant (each week, the downloads double). This makes sense in context: social media sharing often leads to word-of-mouth that doubles reach each cycle. First and second differences were not constant, ruling out linear and quadratic models.

SECTION 7

Comparing the Three Models — Strengths & Limitations

No single function family is "best" — each one excels in certain situations and falls short in others. The table below summarizes when to use each model, what makes it strong, and what its limitations are.

Comparison of the three function families

Feature	Linear	Exponential	Quadratic
When to use	Constant rate of change (e.g., pay per hour, distance at steady speed)	Constant percentage change (e.g., population growth, radioactive decay)	Situation has a max/min or involves area, projectile motion
Graph shape	Straight line	J-curve (growth) or decay curve	Parabola (U or ∩)
Data test	Constant 1st differences	Constant ratios	Constant 2nd differences
Key strength	Simplest model; easy to interpret slope & intercept	Captures rapid growth or decay accurately	Models turning points and symmetry
Key limitation	Cannot model curves, peaks, or acceleration	Grows (or decays) forever — unrealistic long-term for many contexts	Only one turning point; poor for data that levels off

✦ KEY TAKEAWAY

Choosing a function model is like choosing a map projection — every projection distorts something, and the best choice depends on what matters most for your trip. A linear model is the simplest "projection" but can't capture curves. An exponential model nails rapid change but eventually predicts impossibly large (or impossibly small) values. A quadratic model captures a turning point beautifully but can't model sustained growth. Your job is to pick the model that distorts the least for the context at hand.

SECTION 8

Connection to Advanced Modeling

The three families you've learned here — linear, exponential, and quadratic — are the building blocks for more sophisticated models you'll encounter in later courses. Understanding when and why each family works prepares you for the bigger world of data analysis and mathematical modeling.

How today's skills connect to future math courses

What You Know Now	Where It Leads
Choosing between linear, exponential, and quadratic by testing differences/ratios	Regression analysis — using technology to fit the best curve to messy, real-world data
Writing y = mx + b from context	Systems of equations and linear programming for optimization
Writing y = a · bˣ for growth/decay	Logarithmic functions (the inverse of exponentials) and compound interest formulas
Writing y = ax² + bx + c for parabolic situations	Polynomial and rational function modeling; calculus optimization
Justifying your model choice	Residual analysis — measuring how well your model actually fits the data

In AP Statistics or Precalculus, you'll learn to use residual plots and R² values to quantify exactly how well a model fits data. But the intuition you build now — recognizing patterns, testing hypotheses about which family applies, and justifying your reasoning — is the same skill set you'll use at every level.

SECTION 9

Practice Problems

PROBLEM 1 — CONCEPTUAL

A friend says, "The data shows that every time x increases by 1, y gets multiplied by 5. That means the function is quadratic." Explain why your friend is incorrect and identify the correct function family.

PROBLEM 2 — BASIC CALCULATION

A gym charges a $25 sign-up fee and $40 per month. Write a function model for the total cost y after x months. Name the function family and justify your choice.

PROBLEM 3 — INTERMEDIATE

Consider the following data: x = 0, 1, 2, 3, 4 and y = 2, 6, 18, 38, 66. Compute first differences and second differences. Identify the function family and explain your reasoning.

PROBLEM 4 — APPLIED

A biologist observes a bacteria culture in a lab. At time 0 hours, there are 300 bacteria. At 1 hour there are 450, at 2 hours there are 675, at 3 hours there are 1012.5, and at 4 hours there are 1518.75. Determine the function family, write the model, and predict the population at 6 hours.

PROBLEM 5 — CRITICAL THINKING

A farmer has 120 meters of fencing and wants to build a rectangular pen against a barn wall (so only three sides need fencing). Let x represent the width of the pen. Write a function for the area A in terms of x, identify the function family, and explain why this context practically guarantees that family. Then find the dimensions that maximize area.

SUMMARY

Lesson Summary

Building a function model means translating a real-world context into a mathematical equation by identifying the input and output variables, then selecting the function family that best describes the relationship. The three families studied here are linear (y = mx + b) with a constant rate of change, exponential (y = a · bˣ) with a constant multiplicative ratio, and quadratic (y = ax² + bx + c) with a constant second difference. Each family has a distinctive graph shape — a straight line, a J-curve or decay curve, and a parabola, respectively.

To choose the right family, use diagnostic tests on data (first differences, second differences, or consecutive ratios) or look for context clues in verbal descriptions (steady increase → linear; doubles/halves → exponential; maximum height or area → quadratic). Always justify your choice by citing the specific evidence — constant differences, constant ratios, or the real-world behavior — that supports your model. This skill of reasoning about which model fits best is the foundation for all future work in data analysis, statistics, and applied mathematics.