Why Model Evaluation Matters
Explain why a model must be tested instead of trusting its predictions.
Understand the concept
Model evaluation compares a model's outputs with known outcomes using data that was not used to fit the model. It provides evidence about how the model may perform on new cases. No single test guarantees future success.
Business connection
A retailer should compare a sales forecast with actual sales before using it to set inventory levels.
INTERACTIVE VISUAL
Training vs unseen performance
Switch between familiar and unseen performance to understand generalisation.
INTERACTIVE ACTIVITY
Apply the concept
A model scores 99% on training data but only 63% on a suitable unseen test set. What is the best conclusion?