Parsimonious models for multivariate time series

Host Faculty: Engineering
General Subject Area: Statistics and Data Science
Project Level: PhD
View the Research Website
FIND A PROJECT - Engineering

The established framework for modelling multivariate time series is the vector autoregression moving average (VARMA) models. This class of models requires a large set of parameters. We are interested in either sparse structures using statistical learning approaches such as graphical lasso or more parsimonious models such as the recently proposed ZAR models. We are interested in models for both analysis and forecasting. This is a broad area of research: the specific project will depend on the student's interests.


Supervisor: Marco Reale

Does the project come with funding


Final date for receiving applications



Time series analysis; statistical learning; graphical modelling