Stats

Question 1

Mention four reasons for multi-level models

Answer 1

1) Adjust estimates for repeat sampling.
2) Adjust estimates for imbalance sampling.
3) To study variation.
4) To avoid averaging.

Question 2

Mention areas where unsupervised learning can be used.

Answer 2

Cluster analysis: Finding a degree of similarity among the data.

Dimensionalty reduction: estmate a lower dimensional space of the data.

Association rules in the data.

In general it is about estimation of the joint density of X, in contrast to supervised learning where the conditional density of P(Y|X) is edtimated.