Rsq.glm: R-squared measures for binomial GLMs

In binomTools: Performing diagnostics on binomial regression models

Description

This function computes the R-squared measures for binomial GLMs proposed by Tjur (2010) "Coefficients of determination in logistic regression models - a new proposal: The coefficient of discrimination".

Usage

## S3 method for class 'glm'
Rsq(object, ...)

## S3 method for class 'Rsq'
print(x, digits = getOption("digits"), ...)

## S3 method for class 'Rsq'
plot(x, which=c("hist", "ecdf", "ROC"), ...)

Arguments

object	a binomial glm object
x	an Rsq object
which	the desired plot: histograms, empirical cumulative distribution functions or ROC (receiver operating characteristic) curve
digits	the desired number of printed digits
...	currently not used

Details

The plot method has the following options

"hist"

Two histograms with ten bins of the fitted probabilities are plottet on top of each other; the upper one for y = 0 and the lower one for y = 1.

"ecdf"

Two ecdf curves; one for y = 0 and one for y = 1

"ROC"

The (empirical) ROC curve

Value

Rsq.glm returns an object of class Rsq. The plot and print methods returns the Rsq objects invisibly.

Author(s)

Rune Haubo B Christensen

References

Tjur, T. (2009) Coefficients of determination in logistic regression models - a new proposal: The coefficient of discrimination. The American Statistician, 63(4), 366-372.

See Also

A HLtest (Hosmer and Lemeshow test) method exists for Rsq objects.

Examples

## Lifted from example(predict.glm):
ldose <- rep(0:5, 2)
numdead <- c(1, 4, 9, 13, 18, 20, 0, 2, 6, 10, 12, 16)
sex <- factor(rep(c("M", "F"), c(6, 6)))
SF <- cbind(numdead, numalive=20-numdead)
budworm.lg <- glm(SF ~ sex*ldose, family=binomial)
## summary(budworm.lg)

(Rsq.budworm <- Rsq(budworm.lg))

plot(Rsq.budworm, "hist") ## or simply 'plot(Rsq.budworm)'
plot(Rsq.budworm, "ecdf")
plot(Rsq.budworm, "ROC")

Example output

R-square measures and the coefficient of discrimination, 'D':

     R2mod     R2res    R2cor         D
 0.4300889 0.4151594 0.415289 0.4226242

Number of binomial observations:  12
Number of binary observation:  240
Average group size:  20 

binomTools documentation built on May 2, 2019, 6:53 p.m.