Take the weights apart

Year 12 ยท Student investigation

Does a neat decomposition preserve the behaviour we care about?

Keep an individual record even when you investigate in pairs. Record the exact settings, source and data split.

Predict before seeing the answer

Will keeping all 10 components with zero edit reproduce the original model? Could a small matrix error hide a large group error?

Two comparisons

What did you change? What stayed the same? What did you notice?

Challenge your first explanation

Check exact reconstruction at rank 10 and edit 0. Compare ranks 1, 5 and 10, then edit the leading component. Inspect Kiki and other-query denominators before changing the assumed population share.

Make and explain

A paired model-edit report with rank, edit, group loss, exact answers and population assumptions.

Limit the claim

This activity can show _ . It cannot tell us _ . A new case I would test is ____ .

Use fictional data. Download or print your work before leaving.