Presented at the 2007 Japanese Joint Statistical Meeting, 6-9 September, 2007, Kobe University

 

Rotation in Inference and Description

YASUHARU OKAMOTO

 

Introduction

Rotation in inference from and description of data are discussed in this paper.  Fig. 1 shows that data are generated by latent variables and information contained in the data is represented by components, the number of which is less than the dimensionality of the data.  Both in inference and in description, we can determine the optimum spaces with respect to the data, but to identify the axes we need some criteria  outside of statistics.  In identifying the axes, which have substantial meanings, rotation is used.  First, consider rotation in inferencec<To read the continuance, click here (pdf file)>


For more detail, click here (PDF file)

 

Correspondence may be addressed to Y. Okamoto (e-mail: okamotoy@fc.jwu.ac.jp)

 

Up