Canonical Analysis: Ranks, Ratios and Fits

Albers, Casper; Gower, John
April 2014
Journal of Classification;Apr2014, Vol. 31 Issue 1, p2
Academic Journal
Measurements of p variables for n samples are collected into a n× p matrix X, where the samples belong to one of k groups. The group means are separated by Mahalanobis distances. CVA optimally represents the group means of X in an r-dimensional space. This can be done by maximizing a ratio criterion (basically one- dimensional) or, more flexibly, by minimizing a rank-constrained least-squares fitting criterion (which is not confined to being one-dimensional but depends on defining an appropriate Mahalanobis metric). In modern n < p problems, where W is not of full rank, the ratio criterion is shown not to be coherent but the fit criterion, with an attention to associated metrics, readily generalizes. In this context we give a unified generalization of CVA, introducing two metrics, one in the range space of W and the other in the null space of W, that have links with Mahalanobis distance. This generalization is computationally efficient, since it requires only the spectral decomposition of a n× n matrix.


Related Articles

  • Variable Selection in High-Dimension with Random Designs and Orthogonal Matching Pursuit. Joseph, Antony // Journal of Machine Learning Research;Jul2013, Vol. 14 Issue 7, p1771 

    The performance of orthogonal matching pursuit (OMP) for variable selection is analyzed for random designs. When contrasted with the deterministic case, since the performance is here measured after averaging over the distribution of the design matrix, one can have far less stringent sparsity...

  • Role of Discriminantly Separable Polynomials in Integrable Dynamical Systems. Dragović, Vladimir; Kukić, Katarina // AIP Conference Proceedings;2014, Vol. 1634, p3 

    Discriminantly separable polynomials of degree two in each of the three variables are considered. Those polynomials are by definition polynomials which discriminants are factorized as the products of the polynomials in one variable. Motivating example for introducing such polynomials is the...

  • The EBIC and a sequential procedure for feature selection in interactive linear models with high-dimensional data. He, Yawei; Chen, Zehua // Annals of the Institute of Statistical Mathematics;Feb2016, Vol. 68 Issue 1, p155 

    High-dimensional data arises in many important scientific fields. The analysis of high-dimensional data poses great challenges to statisticians. In high-dimensional data, the relationship among the variables is complex. It involves main effects as well as interaction effects of the covariates....

  • Using the NOIR Matrix. Lackey, Jeff // Quality Progress;Jul2003, Vol. 36 Issue 7, p104 

    Provides some insights into the use of the NOIR matrix to remember four different types of statistical variables. Comparison of the types of variables; Type of measurement scale; Allowed mathematical operations; Allowed math transformations; Types of statistical calculations.

  • Summarising categorical variables. Campbell, Malcolm // African Journal of Midwifery & Women's Health;Apr-Jun2015, Vol. 9 Issue 2, p53 

    No abstract available.

  • Effects of socio-economic household characteristics on traditional knowledge and usage of wild yams and medicinal plants in the Mahafaly region of south-western Madagascar. Andriamparany, Jessica N.; Brinkmann, Katja; Jeannoda, Vololoniaina; Buerkert, Andreas // Journal of Ethnobiology & Ethnomedicine;2014, Vol. 10 Issue 1, p1 

    Background Rural households in the Mahafaly region of semi-arid SW Madagascar strongly depend on the exploitation of natural resources for their basic needs and income regeneration. An overuse of such resources threatens the natural environment and people's livelihood. Our study focuses on the...

  • Gestational Alcohol Exposure and Other Factors Associated With Continued Teenage Drinking. Cornelius, Marie D.; Goldschmidt, Lidush; Day, Nancy L. // Health Education & Behavior;Aug2016, Vol. 43 Issue 4, p428 

    Purpose. A longitudinal cohort of adolescents who initiated drinking before age 15 were studied to determine which factors distinguished between early initiators who continued to drink (persisters) from those who stopped drinking (desisters). There were 308 early initiators in the total sample...

  • TOURIST ACTIVITY OF PENSIONERS IN POLAND IN 2009: A STATISTICAL ANALYSIS. Bak, Iwona // Folia Oeconomica Stetinensia;2011, Issue 2, p155 

    The purpose of the article is the analysis of transitory pensioners touristics departures. The research was carried out basing on a representative research result of individual transitory domestic departures in 2009 where pensioners participated in. As investigative instrument the autor chose...

  • Solubility of Systems of Quadratic Forms. Martin, Greg // Bulletin of the London Mathematical Society;1997, Vol. 29 Issue 4, p385 

    It has been known since the last century that a single quadratic form in at least five variables has a nontrivial zero in any p-adic field, but the analogous question for systems of quadratic forms remains unanswered. It is plausible that the number of variables required for solubility of a...


Read the Article


Sorry, but this item is not currently available from your library.

Try another library?
Sign out of this library

Other Topics