Stat 399: Transposable data, Professor Art Owen
This web page is a repository for handouts for Stat 399.
(taught Spring 1999/2000)
This course looks into statistical problems raised
by search engines, recommender systems, and
The common element in these problems is that the data
are given as a matrix with two equally valid interpretations:
rows=observations, columns=variables, and vice versa.
We meet Wednesday at 3:15 in Sequoia Hall 200.
Related research paper (plaid modeling)
A course reader is now available from Stanford
University Bookstore. It costs $27.50 of
which $8.00 covers copyrights for various articles
and book parts copied therein.
Term Document Matrices:
Non-negative matrix decomp (Kristofer Jennings' talk)
Semi-definite matrix decomp (Tao Jiang's talk)
Dyadic Data (Lingyu Chen's talk)
Fuzzy Clustering (Jorge Picazo's talk)
Correspondence analysis (Ji Zhu's talk)
Rasch Models (Phil Beineke's talk)
Data David Marimont mentioned.
of the Nature issue with the Chromosome 21 article
Independent Component Analysis (Josh Stuart's talk)
SVD based clustering (Alan Gous's talk)
Plaid models (Textual part of Art Owen's talk)
More useful links
Latent Class (Rasch) Model
Weigend, Weiner, and Pedersen's Exploiting Hierarchy
More of Andreas Weigend's