CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Semi parametric mixtures of generalised linear models Authors:  Sollie Millard - University of Pretoria (South Africa) [presenting]
Frans Kanfer - University of Pretoria (South Africa)
Mohammad Arashi - Shahrood University of Technology (Iran)
Abstract: Mixtures of generalised linear models and an extension to a semi-parametric mixture setting are considered. The link function is replaced by a non-parametric estimate thereof. This approach allows for more flexibility since the non-parametric link function gives access to a larger subset of distributions in the exponential family, whilst retaining much of the structure of a generalised linear model. The performance of the proposed procedure is evaluated through a simulation study considering various input settings. An industry case study using a semi-parametric alternative to mixtures of logistic regressions is also presented.