Freeze the split on the How to Train Models track. Split before any fitting, including scalers and vocabulary builds. Time-based or user-based splits when random splits would leak.
This lesson assumes you already worked through Choose the objective first.
The idea in practice
Save the row ids of train, validation, and test. Refuse a run that does not load those ids.
A concrete check
goal = {
'track': 'How to Train Models',
'lesson': 'Freeze the split',
}
checks = [
'input available at decision time',
'score matches the real decision',
'failure case written down',
]
print(goal['lesson'])
for item in checks:
print('-', item)
Run the sketch locally if you have Python. The printout is a reminder of the checks, not a trained model. Replace the strings with the real inputs from your own example before you treat it as a design.
What usually goes wrong
Resplitting until the score looks better. When this happens, stop adding parameters or tools. Fix the check, the data, or the permission, then run the same example again.
What to write down
- The input you are allowed to use at decision time.
- The output and the score or pass rule.
- One failure you will test on purpose.
- What you will not claim the system can do.
Practice
Write the split rule and the file where the ids live.
Self-check
- Say Freeze the split in one sentence that mentions an input and an output.
- Name the failure mode in this lesson and the check that would catch it.
Done when: you can explain this lesson without the page open, and you have a written failure case.