Assessing Calibration of Logistic Regression Models: Beyond the Hosmer-Lemeshow Goodness-of-Fit Test

Lieu: CNAM (21.1.13)
Date et Heure de début: 16-02-2018

 The Hosmer-Lemeshow Goodness of Fit Test has become the standard method for assessing calibration for binary logistic regression models.  However, this method breaks down when models are developed from extremely large data sets.  In this talk I will discuss options for assessing model performance in these situations including modifying the number of groups used and a new method called “Calibration Belts”.

Organise: MSDMA
Contact: Avner Bar-Hen