Entropy correlation coefficient for measuring predictive power of generalized linear models
By considering properties of generalized linear models (GLMs), a correlation coefficient, which is referred to as the entropy multiple correlation coefficient, is proposed as a predictive power measure of GLMs. This correlation coefficient is a measure of linearity of response variables and the canonical parameters, and can be viewed as the proportion of reduced uncertainty of response variables by explanatory variables.
Year of publication: |
2007
|
---|---|
Authors: | Eshima, Nobuoki ; Tabata, Minoru |
Published in: |
Statistics & Probability Letters. - Elsevier, ISSN 0167-7152. - Vol. 77.2007, 6, p. 588-593
|
Publisher: |
Elsevier |
Keywords: | Entropy Entropy multiple correlation coefficient Generalized linear model Kullback-Leibler information |
Saved in:
Saved in favorites
Similar items by person
-
Entropy coefficient of determination for generalized linear models
Eshima, Nobuoki, (2010)
-
Eshima, Nobuoki, (2011)
-
Tabata, Minoru, (2002)
- More ...