Define the prediction problem
A prediction is a guess about a target using information available at a stated decision time. Define the task before choosing a model.
One row is one student. At the end of Week 4, inputs are information genuinely available at that time; the target is later withdrawal. A feature is an input measurement or derived summary. A label is the known target for a historical example. A classifier predicts a category such as Withdraw/Stay; predicting a numerical quantity is a different task.
ID is retained for tracking and joins, not included as an arbitrary numerical distance feature. Set membership (train/test/new) and later outcome must not be predictors. Historical outcomes may be training labels and test outcomes may score fixed predictions, but neither role permits leaking an unavailable outcome into inputs.
Sources: W4 PDF pp. 2-42; L4 Tasks 1-7. All fixed worked values here were recalculated from the supplied examples. These toy rules are teaching devices, not deployable student-support recommendations.
The supplied split tells you which records teach the rules and which records only receive predictions. Keep those roles separate.
Practice problems
Supplementary study questions on the lecture concepts — not verbatim exam questions. Each answer counts toward spaced review.
1. Conceptual understanding
Definitions and distinctions.
What should be defined before building a predictor?
For an end-of-week-4 withdrawal prediction, which is a valid boundary?
2. Application
Apply the concept to a case or calculation.
For predicting later withdrawal, which field is the target?
A week-4 prediction uses week-5 logins. What is the problem?
3. Interpretation and limitations
Evaluate assumptions and justify conclusions.
If one prediction is made per student, what should the prediction unit be?
Which input could be valid if recorded and available before week 4?
Explain it in writing
Self-checked, not marked. Write a few sentences, then compare with the example answer. Your writing is saved as you type.
Show an example answer
It sets which information was available. Without it, later measurements may leak into inputs and make evaluation unlike the real use case.
Check: did you answer the question, use evidence or working, and avoid claiming more than the evidence supports?
Show an example answer
Features are available inputs used to make the prediction. The target is the outcome being predicted. A target or answer-revealing field must not be slipped into the inputs.
Check: did you answer the question, use evidence or working, and avoid claiming more than the evidence supports?