Aproksymacja modelu regresji logistycznej Firtha za pomocą ważenia obserwacji
Firth’s Logistic Regression Approximation by Weighting Observations
Author(s): Kamil FijorekSubject(s): Economy
Published by: Wydawnictwo Uniwersytetu Ekonomicznego w Krakowie
Keywords: logistic regression; bias reduction; complete separation; approximate estimation
Summary/Abstract: Firth’s approach to a logistic regression is presented from the perspective of weighted data points. Hidden Logistic Model is reformulated accordingly and two approximations of Firth’s procedure are introduced. A simulation study was conducted to investigate and compare the quality of the approximations.
Journal: Zeszyty Naukowe Uniwersytetu Ekonomicznego w Krakowie
- Issue Year: 923/2013
- Issue No: 23
- Page Range: 87-98
- Page Count: 12
- Language: Polish