Test a model by comparing its predictions with observed data and checking whether its form, assumptions, domain, and outputs make sense in context. A model is appropriate when its mathematical form suits the relationship and reasonable when its results are realistic within the intended domain.
This is SL 2.6: Modelling skills, common to SL and HL, and represents the test and reflect stage: pose, develop, test, reflect, then apply, revise, or reject the model.
| Test | What to examine | Example conclusion |
|---|---|---|
| Fit to data | Compare predicted and observed values. Smaller differences suggest better fit. | The predictions are close to most observations. |
| Shape | Check whether the graph follows the trend. | Exponential is suitable for increasing percentage growth. |
| Domain | State meaningful input values. | Limit time to the measured interval. |
| Assumptions | Identify fixed or simplified conditions. | Growth rate is assumed constant. |
| Outputs | Reject impossible predictions, such as negative populations. | The result is unrealistic. |
| Prediction range | Distinguish interpolation from extrapolation. | Distant extrapolation is unreliable. |
For example, if a model predicts customers when are observed, the percentage difference is