[Statlist] Invitation to talk "Clusterpath Gaussian Graphical Modeling and Covariance Matrix Estimation" by Andreas Alfons

[Carolin Strobl]

Dear colleagues, 

on Thursday November 14, 10:30h
in room AND-3-46, Andreasstrasse 15, Zürich Oerlikon
and in Zoom (link see below)

Andreas Alfons, Associate Professor at the Erasmus University Rotterdam (https://personal.eur.nl/alfons/), will give a presentation on "Clusterpath Gaussian Graphical Modeling and Covariance Matrix Estimation" (abstract see below).

We are looking forward to your participation, no registration is necessary. 

Best regards, 
Carolin Strobl 

#########
Zoom Link:
#########

Thema: Vortrag Andreas Alfons
Zeit: 14.Nov. 2024 10:30 AM Zürich

Beitreten Zoom Meeting
https://uzh.zoom.us/j/68243424365?pwd=pUkLpflawva8DGueizR8WGzb9SHgqF.1

Meeting-ID: 682 4342 4365
Kenncode: 328586


#######
Abstract:
#######

Graphical models serve as effective tools for visualizing conditional
dependencies between variables. However, as the number of variables grows,
interpretation becomes increasingly difficult, and estimation uncertainty
increases due to the large number of parameters relative to the number of
observations. To address these challenges, we introduce the clusterpath
estimator of the Gaussian graphical model (CGGM) that encourages variable
clustering in the graphical model in a data-driven way. Through the use of a
clusterpath penalty, we group variables together, which in turn results in a
block-structured precision matrix whose block structure remains preserved in
the covariance matrix. We present a computationally efficient implementation
of the CGGM estimator by using a cyclic block coordinate descent algorithm.
In simulations, we show that CGGM not only matches, but oftentimes outperforms
state-of-the-art methods for variable clustering in graphical models. In
addition, we discuss estimation of a block-structured covariance matrix based
on the CGGM algorithm. Among a diverse collection of empirical applications, we
demonstrate CGGM’s practical advantages and versatility on data from a survey
using psychometric scales.

----------
Universität Zürich
Psychologisches Institut
Prof. Dr. Carolin Strobl
Lehrstuhl für Psychologische Methodenlehre, Evaluation und Statistik
Binzmühlestrasse 14, Box 27
CH-8050 Zürich

carolin.strobl at psychologie.uzh.ch
www.psychologie.uzh.ch/methoden