Hello
In assessing the adequacy of predictive models e.g. from logistic regression, the 'area under the ROC curve' is often used. However the Brier score and various forms of R-sq (eg Nagelkerke's) are also used.
Presumable there is some relation between these. Before I attempt to derive it I wondered if anyone had a suitable reference.
Phil McShane
Senior Healthcare statistician
Paediatric Epidemiology Group
Unit for Epidemiology and Biostatistics
8.49 Worsley Building
Clarendon Way
Leeds
0113 343 8304
07791492029
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|