Research in Nonlinear and Nonstationary Time Series Models

The main thrust in the area of nonlinear time series in the early 80's came from the time series group of Manchester area. The people who contributed mostly to this area are M B Priestley (State Dependent Models), T Subba Rao (Bilinear Time Series Models), H Tong (Threshold Models), Valerie Haggan and T Ozaki (Amplitude Frequency Dependent Models). Bilinear Models were developed at the same time by Professor Clive Granger who was awarded Nobel Prize in Economics in 2003. The development of statistical tests for linearity of time series based on Higher Order Cumulant Spectra were first developed by T Subba Rao and M M Gabr in the univariate case, then have been extended by several other people in recent years, and are now widely used in economics and signal processing. We are now developing methods to deal with spatio-temporal data with strong applications to global temperatures and modelling the relationship between temperatures and ozone concentrations. The methods can also be applied to microarray data which we are now pursuing very actively.

Academic contact:

Professor Tata Subba Rao, Tel: +44 (0)161 30 63664, E-mail: Tata.Subbarao (@manchester.ac.uk)