The proportion of explained variation (R2) is frequently used in the general linear model but in logistic regression no standard definition of R2 exists.
We present a SAS macro (evlogist - based on Ref. 1), which calculates two R2-measures based on Pearson and on deviance residuals for the logistic regression. Also adjusted versions for both measures are given, which should prevent the inflation of R2 in small samples.
For the sums-of-squares version of R2 this relative measure is complemented (Ref. 3) by two related absolute measures, MSE and MST. Furthermore we offer an alternative description of explained variation which is interpretable on the scale of prediction proportions rather than on the scale of variances as is R2. All relative and absolute measures of explained variation are available in shrunk versions particularly recommended for small samples and/or many prognostic factors.
Referenzen:
Mittlböck, M., Schemper, M. (2002): "Explained Variation for Logistic Regression - Small Sample Adjustments, Confidence Intervals and Predictive Precision", Biometrical Journal 44, 1 - 10
Mittlböck, M., Schemper, M. (1999): "Computing measures of explained variation for logistic regression models", Computer Methods and Programs in Biomedicine 58, 17 - 24
Mittlböck, M., Schemper, M. (1996): "Explained variation for logistic regression", Statistics in Medicine 15, 1987 - 1997
Beide Makros sind in der Version 6.12 geschrieben
(eine Einzelversion beider Makros ist in Entwicklung):
Download: SAS macro evlogist for reference 1 and 2 |
Download: SAS macro evl_shr for reference 3 |