Skip to main content
eScholarship
Open Access Publications from the University of California

Majorizing Logit Loss Functions for the Multidimensional Representation of Categorical Data

Abstract

The data we study are measurements of n objects on each of m variables. This corresponds with the spreadsheet format in many software packages and with the data-frame in S/R. In our setup all data are categorical, which means each variable maps the observations into a finite number of categories. Categories can be a finite subset of the reals, a finite ordered set, the set {0,1}, or just an arbitrary finite set.

Main Content
Current View