Prof. Lenny Smith  TW1 11.01A

Primarily for MSc Statistics, MSc Statistics (Research), MSc Statistics (Financial Statistics) and MSc Statistics (Financial Statistics) (Research), MSc Applicable Mathematics, MSc Econometrics and Mathematical Economics.

ST422 Time Series.

An introduction to the analysis of actual time series observations of real-world processes. The course casts both modern nonlinear methods and more traditional linear methods in a geometric approach. It introduces the properties of nonlinear mathematical models, covers chaos and the dynamics of uncertainty, and demonstrates the fundamental limitations in applied analysis which arise from model inadequacy.

The student will leave with a toolkit for the analysis and modelling of real data, with insights into how to evaluate which methods to employ (linear/non-linear, deterministic/stochastic) in a given problem, how to interpret the results in context, and how to avoid over interpreting nice theorems in practical circumstances. Concrete applications in economics (price time series, electricity demand, energy futures) and environment (weather, climate) as well as analytically tractable illustration from mathematics are considered.

Lectures: 20 LT, two ST. Computer Workshops: 10 LT.

K Beven, Environmental Modelling: An uncertain Future? Routledge (2009); H Kantz & T Schreiber, Non-linear Time Series Analysis; E Ott, T Sauer & J A Yorke (Eds), Coping with Chaos: Analysis of Chaotic Data and The Exploitation of Chaotic Systems; E Ott, Chaos in Dynamical Systems; R Tsay, Analysis of Financial Time Series; L.A. Smith, Chaos: A Very Short Introduction. Oxford University Press (2007)

Two-hour written examination in ST: 70%; project: 30%.