Publication Results

Code: 63/2017
Title: Non-parametric mixed-effects models for unsupervised classification of Italian schools
Date: Thursday 23rd November 2017
Author(s) : Masci, C.; Paganoni, A.m.; Ieva, F.
Download link:
Abstract: This paper proposes an EM algorithm for non-parametric mixed-effects models (NPEM algorithm) and applies it to the National Institute for the Educational Evaluation of Instruction and Training (INVALSI) data of 2013/2014 as a tool for unsupervised clustering of Italian schools. The main novelties introduced by NPEM algorithm, when applied to hierarchical data, are twofold: first NPEM allows the covariates to be group specific; second, it assumes the random effects to be distributed according to a discrete distribution P* with an (a priori) unknown number of support points. In doing so, it induces an automatic clustering of the grouping factor at higher level of hierarchy, enabling the identification of latent groups of schools that differ in their effect on student achievements. The clustering may then be exploited through the use of school level features.

This report, or a modified version of it, has been also submitted to, or published on
Published: Masci, C., Paganoni, A. M., & Ieva, F. (2019). Semiparametric mixed effects models for unsupervised classification of Italian schools. Journal of the Royal Statistical Society: Series A (Statistics in Society), 182(4), 1313-1342.