Project Step: Algorithm Candidate Plan
Unit ID: AMLA-M01-U05 Estimated active time: 35-50 minutes
Classroom explanation
Now bring the module together. You are not fitting models yet. You are planning which models deserve attention.
This is a serious modelling skill. Good model builders reject many ideas before spending compute or trust on them.
Required plan
Create a short candidate plan with these sections:
- Problem statement.
- Data shape.
- Decision need.
- Baseline ladder.
- Candidate models to try.
- Models to reject for now.
- Metric and validation plan.
- Limitations.
Candidate reasoning
For every model you include, explain why it is reasonable.
For every model you reject, explain why it is not a good first choice. Rejection can be based on small data, too many missing values, poor interpretability fit, runtime risk, metric mismatch, or lack of evidence.
Practice
Draft your candidate plan in the algorithm comparison report template. Keep it to one page.
Takeaway
The first deliverable in an algorithm course is not a fitted model. It is a defensible plan for what deserves to be fitted.
