Friday, February 24
PS2 Poster Session 2 and Refreshments Fri, Feb 24, 5:15 PM - 6:30 PM
Conference Center AB

Iterative Semiparametric Generalized Linear Models (303453)

Ayse Aysin Bombaci Bilgin, Macquarie University 
Timothy Kyng, Macquarie University 
Jun Ma, Macquarie University 
*Busayasachee Puang-Ngern, Macquarie University 

Keywords: Semiparametric Generalized Linear Models, Generalized Linear Models, Semiparametric, Iterative method, GLMs, SP-GLMs

The semiparametric generalized linear models (SP-GLMs) are extensions of generalized linear models (GLMs). They can provide better fit to the data than the classical GLMs due to the additional flexible nonparametric component. SP-GLMs can be used for medical research, actuarial modelling and regression analysis. However, SP-GLMs have computational difficulties. They involve nonparametric component which is difficult to compute. Moreover, imposing constraints to SP-GLMs make them hard to compute. In this poster we outline an iterative method to estimate both the regression coefficients and the nonparametric component simultaneously. This new computational method provides accurate estimations and is able to handle large sample sizes, while the existing estimation methods are restricted to small samples. We will also present a simulation study and an application to a medical data set.