Unbiased Estimation For Logistic Regression
Yükleniyor...
Tarih
2007
Yazarlar
Dergi Başlığı
Dergi ISSN
Cilt Başlığı
Yayıncı
Beykent Üniversitesi
Erişim Hakkı
Özet
The aim of this study was to determine and reduce the problems that have been encountered in sparse data or separation. For this aim, Firth's Modified Score Procedure, which is more superior to Maximum Likelihood Estimation Method, were performed on a data set regarding psychology. Under some circumstances (sparse data or separation), maximum likelihood estimates of parameters are biased or infinite estimates. Using Firth's Modified Score Procedure or (Firth Type Estimates) may be thought as an originally approach which eliminates or shrinks the first order bias. As a result, it has been suggested in this article that Firth Type Estimates be a reliable method for eliminating separation.
Açıklama
Anahtar Kelimeler
Bias shrinking, Firth Type Estimation, Sparse data, Separation
Kaynak
WoS Q Değeri
Scopus Q Değeri
Cilt
Sayı
Künye
Journal of Science and Technology 1 (2), 2007, 245-251