Complex time series modelling and forecasting
The Statistics and Probability group at York has built up extensive expertise in time series related advances.
Much of this is applied to economics, but also to scientific fields such as neurosciences and biology.
As described at greater length in the Large-scale data analytics section, a huge volume of interrelated data needs to be accommodated. Challenges also arise from the non-stationary behaviour exhibited by many time series. In many instances, such behaviour may be coupled with missingness due to, for example, equipment malfunction or human error.
Although many practitioners still choose to treat such time series data as regularly sampled, stationary data in order to make use of traditionally available tools for modelling and forecasting, the results have often been proved to be misleading. Our group publishes work directed precisely at closing these gaps.
Away from economics, new and unusual application areas continue to open up and motivate novel statistical methods. For example, members of the group are involved in projects to analyse rhythmic patterns in brain activity. Others work on the dynamics of opinion-forming and decision-making in groups of experts.