GLMM Based Clustering of Multivariate Mixed Type Longitudinal Data

Abstract

Several GLMMs for longitudinal data of a mixed type are joined together. Then, a mixture of such models is proposed to classify units into groups differing in evolution patterns.

Data from The European Union Statistics on Income and Living Conditions database (EU-SILC) are analysed.

Keywords

• Multivariate longitudinal data
• GLMM
• Model-based clustering
• Sparse finite mixture