Abstract
The canonical correlation technique is a simple and useful method to describe the correlation structure between two sets of variables. It is a generalization of the concept of multiple correlation and successively maximizes the correlation between appropriate pairs of linear combinations of the variables of the two sets. The method can be viewed as a dimension reduction technique in that it represents the correlation structure between two sets of variables in terms of a smaller number of canonical correlations. In this article, we give a basic description of the canonical correlations technique and illustrate it with an application.
Original language | English (US) |
---|---|
Title of host publication | International Encyclopedia of Education |
Publisher | Elsevier Ltd |
Pages | 52-57 |
Number of pages | 6 |
ISBN (Print) | 9780080448947 |
DOIs | |
State | Published - 2010 |
Externally published | Yes |
Keywords
- Canonical variables
- Correlation
- Correlation coefficient
- Dimension reduction
- Eigen vectors
- Eigenvalues
- Linear combination
- Multiple correlation
- Multivariate analysis