Advanced level · Module 6: Machine Learning Capstone · Lesson 3 of 5
Modelling and Interpretation
Develop forecasting models that beat the baselines and explain their behaviour.
In this lesson you will
- Build and tune forecasting models
- Interpret and document their behaviour
Key concepts
feature-based modelstuning with backtestsinterpretationuncertainty intervals
The lesson
Develop feature-based models, tune them using the backtesting framework and keep the test period untouched until the end.
Interpret the final model with permutation importance and example forecasts, and add calibrated prediction intervals.
Check where the model fails — specific markets or periods — and record it for the model card.
Step by step
- Develop and tune candidate models.
- Evaluate on the final test period.
- Interpret the chosen model.
- Add and check prediction intervals.
Practice activity
Complete milestone 3: the final model with interpretation and uncertainty.
Expected result: A model that meets the success criteria where possible, with honest reporting where it does not.
Check your understanding
-
Where should failures be recorded?
Show answer
In the model card's limitations section.
-
When is the final test period used?
Show answer
Once, after all modelling choices are made.
Taught live by an instructor — online or in person. Free course lessons are for self-study and do not lead to a certificate — see registration options.