Interactive threshold experiment
This deliberately small synthetic dataset contains fictional sensor readings and example yes/no outcomes. Move the decision threshold to see how the model makes different mistakes. This is a rule-based demonstration of classification metrics, not a trained machine-learning model. All calculations run in your browser.
| ID | Sensor score | Actual: needs attention? | Prediction? |
|---|
Challenge and next step
- Test thresholds 40, 55, 70 and 85. Write down the four counts for each.
- Decide what you would prioritise if a missed positive had a bigger impact than a false alarm.
- Explain why this experiment is not enough to justify deploying a real system.