Skip to content
SPM Tuition
Lesson · Mathematics

Does the model work outside the data?

Your model fits the table perfectly, and then it predicts a candle with negative length.

A model is built from data in a certain range, and it may fail outside that range. Before you trust a prediction, check whether the value is possible and how far it lies from the data.

This lesson is part of choosing a model from imperfect information. It sharpens the last step of explaining limitations and refining a model.

What do I check before trusting a prediction?

Ask three questions, in this order.

  1. Is the predicted value physically possible?
  2. Is the rule still true at that input?
  3. How far is the input from the data I used?

A “no” to the first two, or a large distance in the third, means the prediction needs a warning.

Worked example: a burning candle

A candle is 20 cm long when lit. Its length L (cm) is measured each hour.

t (hours) 1 2 3 4
L (cm) 18 16 14 12

Model. The length drops 2 cm each hour, so L = 20 − 2t. Check: t = 3 gives 20 − 6 = 14, which matches.

Predict t = 12. L = 20 − 24 = −4 cm. A candle cannot have a negative length, so the prediction fails the first question.

Find the limit. The candle is gone when L = 0, so 20 − 2t = 0 and t = 10. The model holds for 0 ≤ t ≤ 10.

Statement. The model is valid for 0 ≤ t ≤ 10. At t = 12 the candle has burnt out, and the real length is 0 cm.

The data covered 4 hours, and the valid range is 10 hours. Even so, predictions at 8 hours lean on the assumption that the burn rate stays constant.

The mistake that costs marks

The common slip is to report L = −4 without comment. The arithmetic is right, and the answer is impossible.

Step Wrong Right
Calculate L = 20 − 2(12) = −4 L = −4, which is impossible
Check not done a length cannot be negative
Conclusion length is −4 cm the candle has burnt out by t = 10, so L = 0
Range not stated 0 ≤ t ≤ 10

A quick look at the sign of the answer would have caught the slip.

Check yourself

A tank holds 10 litres at the start and fills at 5 litres per minute, so V = 10 + 5t. The tank’s capacity is 60 litres. What is the valid range, and what happens at t = 12?

Answer

The tank is full when 10 + 5t = 60, so 5t = 50 and t = 10. The model is valid for 0 ≤ t ≤ 10.

At t = 12 the model gives V = 10 + 60 = 70 litres, which exceeds the capacity. The real volume is 60 litres, and the extra water overflows.

What to study next

Go on to explaining two defensible models when a question leaves an assumption open. For a graph-based way to compare a model with data, try the graph evidence comparison lab.

The integrated practice set mixes all four skills. For a teacher to work through range checks with you, see online one-to-one Mathematics tuition.

Common questions

What is extrapolation?

It means using a model to predict outside the range of the data it was built from. It is riskier than predicting inside the range, because the pattern may change beyond what was measured.

How can I tell a prediction is impossible?

Check it against the physical situation. A length cannot be negative, a tank cannot hold more than its capacity, and a count of people cannot be a fraction. If the value breaks such a rule, the model is out of range.

Should I write a valid range with the model?

Yes. Stating the range, such as 0 ≤ t ≤ 10, tells the reader where the model can be trusted, and it can earn marks.

Is a prediction inside the data range always safe?

It is safer, but not certain. Real values have small errors, so treat the prediction as an estimate and round sensibly.

If your predictions come out impossible and you only notice at the end, one-to-one Mathematics lessons let a teacher build the range check into your routine on your own questions.

  • Online one-to-one lessons for your child with an experienced teacher.
  • Your first class is a one-hour trial, from RM50. The fee is agreed before you book.
  • Happy with the teacher? Continue with lessons of about 1.5 hours. If not, ask for another teacher.