This is a reminder that there is a Statistics seminar at UMIST tomorrow,
Wednesday 3rd March. Details are given below. The seminar will be followed
by tea and biscuits. All those interested are welcome to attend.
Please visit our web site
http://www.ma.umist.ac.uk
to obtain information about forthcoming seminars.
Wednesday 3rd March, 4.00pm.
Venue: UMIST, Room O10.
Sepaker: Dr Granville Tunnicliffe-Wilson, University of Lancaster
Title: Conditional Independence Modelling of Multivariate Time Series.
Abstract:
Conditional independence modelling can reveal the relationship between a set
of variables, using sample partial correlations in a least squares context.
The method is applied to contemporaneous and lagged variables of a
multivariate time series to identify structural AR and ARIMA models. These
allow simultaneous as well as lagged dependence between the series. Examples
illustrate how models with a relatively sparse parameterisation can be
constructed although a unique identification is not always possible.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|