Instructor: Ernst Linder, M307 Kingsbury Hall, 2-2687, elinder@math.unh.edu

Meeting Times: M/W 6:40 – 8:00 pm, 314 Kingsbury Hall (to be changed)

Textbook (required): Shumway R. H. and Stoffer, D.S. : Time Series Analysis and Its Applications. Springer Verlag, 2000. . ISBN 0-387-98950-1

Software: S-PLUS Version 6. UNH site license allows students to copy the full package (Windows version) onto their own PC. We will frequently hold class in the departmental computer lab in Morse Hall using the Linux version of S-PLUS.

Topics: Time Series is a vast topic and can not possibly covered in depth in a single semester. The first third of the course will be devoted to in-depth coverage of classical time-domain analysis including estimation of trend; autocorrelation functions; ARIMA model identification, estimation and forecasting. During the second third of the course we will introduce some modern topics such as unit-root/fractional differencing, interventions, and state space modeling; ARCH models and categorical time series, concluding with multiple time series analysis via transfer function modeling. The final third will be devoted to frequency domain analysis with discussion of spectral analysis, the fast Fourier transform, filtering, and, if time permits, Wavelet analysis and, possibly, multivariate versions of these.

Style: The course will have an applied focus. We will analyze many real time series drawn from a variety of scientific areas, such as economics, social, environmental, and space sciences. The underlying theory that is required for practice of time series analysis will be introduced at an intuitive level. Mathematical derivations will be kept at a minimum.

Prerequisites: Introductory statistics; some basic knowledge of regression analysis and rudimentary matrix notation and algebra.

Assignments: Weekly homework sets
A team project during the last third of the course. (teams of 1 – 3 students).
Final Exam.

Grade Calculation: Homework 60%, Project 20%, Final 20%