LogisticDx (version 0.1)

logiGOF: Goodness of fit tests for a logistic regression model

Description

Gives 15 commonly employed measures of goodness of fit for a logistic regression model

Usage

logiGOF(x, g = 10)

Arguments

x

A model of class glm

g

No. groups (quantiles) into which to split observations for Hosmer-Lemeshow and modified Hosmer-Lemeshow tests.

Value

• A list of class logiGOF with the following items:
• chiPearCov
  Pearsons chi-square, calculated by covariate group, with $p$ value and interpretation
• chiPearIndiv
  Pearsons chi-square, calculated by individual observation, with $p$ value and interpretation
• chiPearTab
  Pearsons chi-square, calculated by table of covariate patterns by outcome, with $p$ value and interpretation
• OsRo
  Osius & Rojek test of the logistic link, with $p$ value and interpretation
• chiDevCov
  Deviance chi-square, calculated by covariate group, with $p$ value and interpretation
• chiDevIndiv
  Deviance chi-square, calculated by individual observation, with $p$ value and interpretation
• chiDevTab
  Deviance chi-square, calculated by table of covariate patterns by outcome, with $p$ value and interpretation
• covPatTab
  Matrix of covariance patterns, used to calculate above chi-square tests of Pearson residuals and deviance
• HosLem
  Hosmer & Lemeshow goodness of fit test, with g quantile groups,with $p$ value and interpretation
• modHosLem
  modified Hosmer & Lemeshow goodness of fit test, with g quantile groups, with $p$ value and interpretation
• CesHou
  le Cessie, van Houwelingen, Copas & Hosmer unweighted sum of squares test for global goodness of fit, with $p$ value and interpretation
• Stuk
  Stukels test of the appropriateness of the logistic link, with $p$ value and interpretation
• PR2
  Pearsons R^2, correlation of observed outcome with predicted
• ssR2
  Linear regression-like sum-of-squares R^2, using covariate patterns
• llR2
  Log-likelohood based R^2, calculated by covariate group
• ROC
  Area under the Receiver Operating Curve, with 95% CI by method of DeLong

See Also

logiDx

Examples

set.seed(1)
m1 <- genLogiDf(n=100)$model
logiGOF(m1)