Not sure which lesson to do first? Find your level and get a plan →

CS Thinking · Computational Thinking · Grade 6-8

Modeling

Modeling is building a simplified version of a real system that keeps only the details your question depends on.

model output=f(inputs,assumptions)

Quick Answer

Modeling is building a simplified version of a real system that keeps only the details your question depends on. You use it when the real thing is too big, slow, costly or risky to experiment on: the lunch line at noon, a cold spreading through a school, traffic at a junction. You pick the quantities that matter, write a rule for how they change, run the rule to get a prediction, and then compare that prediction with what really happens. A model's answer is only as good as the assumptions built into it.

Why This Matters

A model lets you try a decision before paying for it: add a third lunch server on paper and see whether the line shrinks, or test a new traffic-light timing without causing a jam. Without one you either guess or experiment on the real thing. The risk runs the other way too: a model with an unstated assumption can produce a confident number that is simply wrong.

Intuitive Explanation

Suppose the cafeteria wants to know how long the lunch line gets. About 8 students join it each minute and the servers handle 5 a minute. Names, food choices and shirt colours do not change how many people are waiting, so you leave them out. What is left is one rule: each minute the line grows by 8 minus 5, which is 3. Starting from an empty line, the rule predicts 3, 6, 9, 12 students, and 30 after ten minutes.

Writing that rule was the modeling. Nobody handed it to you; you decided which quantities mattered, how they connect, and what to ignore. Each of those choices is an assumption: arrivals stay at 8 a minute, nobody gives up and leaves. Write the assumptions down, because they are where a model breaks. If the real line at 12:10 holds 18 students instead of 30, the model has still helped: it points you to the arrival rate after the first rush as the assumption to check.

A photo of the line, or a floor plan of the cafeteria, also leaves out detail, but you cannot run either one forward to ask what happens next. That is a representation, not a model. A model takes inputs, applies a rule and produces a prediction you can test. And when a problem already gives you the rule and only asks you to plug in numbers, you are using someone else's model rather than building one.

Think of it like this. A model works like a scale model of a bridge tested in a lab: it keeps the shape and strength that decide whether the bridge holds, and drops the paint and the lampposts. What it shows about loads predicts the real bridge, but only for the features it kept; it can tell you nothing about traffic noise.

Core idea. Every model depends on assumptions, so good models must be checked against real data or known behavior.


When to Use

Reach for modeling when you must predict what a real system will do, or test a change, and no rule has been given: how long a queue gets, how fast a rumour spreads, when a water tank runs dry. Signals include what if, estimate over time, and a situation too large or risky to try for real. If the rule is already written and you only substitute values, you are evaluating. If you only need a simpler picture with no prediction, that is abstraction; running an existing model step by step many times is simulation.

Tip. Ask: Are you deciding which parts of a real situation to keep, and what rule links them, so you can predict its behavior?

How to Recognize It

Check whether you are building the simplified version of a real situation, before any computer runs it.

  1. Is there a real situation, like a lunch line or the fish in a pond, that you want to predict or explain?

    Yes is where modeling starts. If there is no real situation, only a task to finish step by step, such as sorting a list, you need an Algorithm.

  2. Are you choosing quantities to track, such as people arriving per minute, and writing how they change?

    Yes is modeling. If you are only dropping details to make something simpler, with no rule for how things change, it is Abstraction.

  3. Is the model being written down, rather than run forward to see what happens?

    Yes keeps you at modeling. Once a computer repeats the rule over many time steps, day 1, day 2, day 3, you are running a Simulation.

  4. Have the assumptions, like 'nobody leaves the line early', been stated?

    Yes means the model can be judged fairly. If they are missing, list them before trusting any prediction, since every prediction depends on them.

Modeling vs Simulation vs Abstraction vs Algorithm

Modeling uses abstraction, a simulation runs a model, and a model is often turned into an algorithm, so the words get swapped. Ask whether you are building the stand-in, running it, trimming details, or listing steps.

Modeling

Meaning
Modeling builds a simplified version of a real system, with chosen quantities and rules, so you can predict or explain it.
Key test
You choose what to track and how it changes, for a real situation.
Formula
output=f(inputs,assumptions)
Example
Tracking only the fish count, births and catches to predict a pond's fish next year.

Simulation

Meaning
A simulation runs a model on a computer, step by step, to watch what the system would do.
Key test
The model already exists and is being run through time.
Formula
St+1=f(St,P)
Example
Running the pond model year by year for 20 years to see whether the fish die out.

Abstraction

Meaning
Abstraction means ignoring the details that do not matter for the problem at hand.
Key test
Details are dropped, but no rule for change is written.
Example
Describing your route to school by street names only, leaving out the houses.

Algorithm

Meaning
An algorithm is an exact list of steps that solves a problem or finishes a task.
Key test
It gives steps to follow, not a picture of a real system.
Formula
output=f(input)
Example
Finding the largest number in a list: keep the biggest seen so far while checking each item.

Formula & Notation

model output=f(inputs,assumptions)
A computational model defines variables, rules, and assumptions that map system inputs to predicted outputs. Its usefulness depends on fidelity to the relevant aspects of the original system.

Worked Examples

Example 1 — Model a lunch line

Easy

The cafeteria opens at 12:00 with an empty line. About 8 students join the line each minute, and the servers together serve 5 students per minute. Build a model and use it to predict how many students are waiting at 12:06.
Show the steps and the answer
  1. Keep three quantities: arrivals per minute (8), students served per minute (5), and the line length. Leave out names, food choices and payment method.

    The question asks only how many students are waiting, and none of the dropped details changes that count.

  2. Write the update rule: line next minute = line now + 8 - 5.

    Each minute adds the new arrivals and removes the students served, a net change of +3.

  3. Trace the rule from 12:00: 0, 3, 6, 9, 12, 15, 18.

    Six applications of the rule take the line from 12:00 to 12:06.

  4. List the assumptions: 8 arrivals every minute, nobody leaves the line, the servers never slow down.

    These are the points where the prediction can fail, so they belong in the answer next to the number.

Answer: About 18 students at 12:06, assuming 8 arrivals and 5 served every minute.

Takeaway: A model is a rule you chose plus the assumptions behind it; report both, not only the number.

Example 2 — Fixed amount or fixed percent?

Standard

A seedling is measured each day: day 0, 2 cm; day 1, 3 cm; day 2, 4.5 cm; day 3, 6.75 cm. Build a model to predict its height on day 5, and say how far ahead you would trust it.
Show the steps and the answer
  1. Find the day-to-day differences: 1, 1.5 and 2.25 cm.

    Equal differences would call for a rule that adds a fixed amount. These keep growing, so an adding rule would fall further behind each day.

  2. Find the ratios: 3 ÷ 2 = 1.5, 4.5 ÷ 3 = 1.5, 6.75 ÷ 4.5 = 1.5.

    Equal ratios mean the height is multiplied by the same factor every day, so the rule is: next height = height × 1.5.

  3. Run the rule: day 4 = 6.75 × 1.5 = 10.125 cm; day 5 = 10.125 × 1.5 = 15.1875 cm, about 15.2 cm.

    Two more applications of the rule reach day 5.

  4. Push the model to an extreme: day 20 gives 2×1.520≈6650 cm, about 66 m.

    No seedling grows 66 m in three weeks, so the multiply-by-1.5 assumption must fail as the plant matures. Trust the model only a few days past the data.

Answer: About 15.2 cm on day 5, using next height = height × 1.5; trust it only a few days beyond day 3.

Takeaway: Let the data choose the form of the rule (equal differences or equal ratios), then push the model to an extreme to find where it stops making sense.

Example 3 — Find the mistake: a prediction is not a fact

Application

Using the lunch-line model (start empty at 12:00, 8 arrivals and 5 served per minute), Sam writes: "The model says 18 students at 12:06, so there will be exactly 18." At 12:06 the supervisor counts 11. She also recorded the arrivals in each minute: 8, 8, 7, 6, 6, 6. Find the mistake and correct it.
Show the steps and the answer
  1. Compare prediction and count: 18 predicted, 11 counted, a gap of 7.

    A gap that large means at least one assumption in the model did not hold today.

  2. Add the recorded arrivals: 8 + 8 + 7 + 6 + 6 + 6 = 41, against the model's 6 × 8 = 48.

    The model assumed a steady 8 per minute, but arrivals fell after the first rush. The missing 7 students explain the whole gap.

  3. Rerun the rule with the real arrivals: 0, 3, 6, 8, 9, 10, 11.

    Each minute still removes 5: 0+8-5 = 3, 3+8-5 = 6, 6+7-5 = 8, 8+6-5 = 9, 9+6-5 = 10, 10+6-5 = 11, which matches the count.

  4. Rewrite Sam's claim with its assumption attached.

    Sam treated the model's output as certain; an honest prediction says what the number depends on.

Answer: Sam treated the model's output as fact. Corrected: "If 8 students arrive every minute, expect about 18 at 12:06." With the recorded arrivals (41 in six minutes) the same rule gives 11, matching the count.

Takeaway: A model's number is a prediction under its assumptions. When reality disagrees, find the assumption that broke and update the model.

Where You'll See It

  • A weather app predicting tomorrow's rain

    The app runs a simplified model of the atmosphere using temperature, pressure, and wind — it ignores countless tiny details, which is why a forecast is a good estimate, not a guarantee.

  • A board game as a model of a city

    Monopoly models buying property and paying rent but leaves out traffic, weather, and real prices — it keeps the details that make the game work and drops the rest, just like a computational model.

  • A flight simulator for learning to fly

    A simulator models how a plane responds to controls so students can practice safely — it is a focused, simplified version of reality built around the question 'how do I fly?'.

Common Mistakes

  1. The slip: A more detailed model is always a better model.

    Why it's tempting: More detail feels more realistic, so it seems more accurate.

    Instead: A model should keep only the details that matter for the question; extra details add complexity and can obscure the answer without improving it. Good models are deliberately simplified.

  2. The slip: If a model gives a number, that number is the truth.

    Why it's tempting: A precise-looking output feels authoritative, like a fact.

    Instead: Model output is an approximation that depends on its assumptions. You must check it against real data or known behavior, and state the assumptions, before trusting the result.

  3. The slip: Once the rule and the prediction are written down, the model is finished; the assumptions are obvious, so there is no need to list them.

    Why it's tempting: A rule like 'the line grows by 8 minus 5 each minute' seems to speak for itself.

    Instead: Every rule rests on assumptions: the lunch-line rule assumes 8 arrivals and 5 served every minute, and that nobody leaves the line early. Write them next to the answer, because when the model predicts 30 students after ten minutes and the real line holds 18, the list tells you which assumption to check first.


Mini Practice

Try these on your own. Tap Reveal when you want to check.

  1. A tank holds 50 L and a hose adds 8 L every minute. Write a model for the volume after m minutes, and use it to find the volume after 12 minutes.

    Hint: Start amount plus a fixed amount per minute.

  2. A town of 5,000 people grows by 3% each year. Write a model for the population after t years and use it for t=2.

    Hint: Growing by 3% means multiplying by the same number every year.

  3. You model the walk to school as time = distance ÷ speed. The distance is 1.2 km and your speed is 4 km/h. Predict the time in minutes and name one assumption.

    Hint: Find the time in hours first, then convert to minutes.

  4. To model how long a pizza delivery takes, which two of these details matter most: distance, traffic, the driver's shirt colour, the pizza topping?

    Hint: Keep only the details that change the time.

  5. A problem gives you C=15n+40 for the cost of n tickets and asks for the cost of 10 tickets. Is this modeling or solving an equation?

    Hint: Who built the relationship: you, or the problem?

  6. A model predicts the lunch line will be 23.6 students long at noon. A student says: 'So exactly 23.6 students will be in line.' Correct the reasoning.

    Hint: Is a model's output exact truth or an estimate?

Want the full set?

50 practice questions for this concept — free to try, every one with a complete worked solution showing the why, not just the answer.

Frequently Asked Questions

If every model leaves things out, how can it be useful?

Because leaving things out is what makes it usable. The statistician George Box put it this way: all models are wrong, but some are useful. A subway map is wrong about distances yet perfect for working out where to change trains. The test of a model is not whether it is complete, but whether it answers your question well enough.

Can two different models of the same thing both be right?

Yes, if they answer different questions. A model of a school for planning fire drills tracks doors, hallways and how fast people walk. A model of the same school for planning the budget tracks teachers, students and costs. Neither is wrong; each keeps what its question needs. Trouble starts only when you use a model to answer a question it was not built for.

Is a spreadsheet a model?

It can be. A spreadsheet where you type how many tickets you expect to sell and the price, and formulas work out the profit, is a model of your event: inputs, rules and a prediction. Changing an input to see what happens, such as dropping the price from $5 to $4, is asking the model a 'what if' question, one of the main reasons to build one.

Learning Path

Modeling

You are here

Next →

Nothing after this yet
You need Abstraction (choosing which details to drop) and Simulation (running a rule step by step) first. Modeling adds choosing the rule itself and testing it against real observations. With it you can compare designs before building them, such as two or three lunch servers, and build spreadsheet models of a budget, a game economy or a school fundraiser.

See Also