Sofia Gvaladze

Methodology of Educational Research
Faculty of Psychology and Educational Sciences
KU Leuven

Phone: +32 16 30 10 43
Email Sopiko Gvaladze


Capturing time-varying multivariate dynamics through principal component analysis based methods
Nowadays improvements in technology allow to collect multivariate time-series on multiple response channels. Understanding the multivariate dynamics present in these data and how they change across time is not easy in case the number of variables grows larger. In this case, dynamic dimension reduction based methods may be a great help for capturing the major features of the data. These methods may pinpoint how variables covary and influence one another across time, and how these dynamics are affected by interventions or important events. When studying multiple individuals, multiset extensions of these methods could shed light on which of these dynamical aspects are shared by different individuals and what makes them unique. The aim of this project is to develop such methods, based on principal component analysis, and to compare their performance to existing techniques. The new methods will be applied to empirical data and disseminated to substantive researchers by, amongst others, building easy-to-use software.

Eva Ceulemans, Francis Tuerlinckx, Peter Kuppens

Financed by

2016 – 2020