Time Series Analysis

Announcements.
The file arma.code contains the ARMA code that I have been using in class. You can define these in your own S directory functions by using the S function `source' to load the contents.
Lecture notes.
  1. Overview.
  2. Stationary processes.
  3. Linear and ARMA processes.
  4. Covariance functions of ARMA processes.
  5. Covariance functions (continued).
  6. Introduction to state-space models.
  7. State-space ARMA models.
  8. Kalman filtering.
  9. Applications of Kalman filtering.
  10. Computing the Kalman filter and the Cholesky factorization.
  11. Hilbert spaces.
  12. Hilbert spaces of random variables.
  13. Time series in the frequency domain.
  14. The spectral representation of a time series.
  15. Applications of the spectral representation.
  16. The Wold decomposition.
  17. Estimating trend.
  18. Estimators of the covariances.
  19. Estimators for autoregressions.
  20. Nonlinear least squares.
    A supplemental file varstar.S contains S code which illustrates Gauss-Newton estimation for the variable star data .
  21. Maximum likelihood for ARMA models.
    An S file was used in class for our analysis of the CPI data .
  22. Basic diagnostics for ARMA models.
    An S file guiding our analysis of the unemployment data was used in class.
  23. Model selection.
  24. Model selection via the AIC.
  25. Introduction to information theory.
  26. Model selection via the Minimum Description Length.
  27. Estimation in the frequency domain.
  28. Spectral estimation.
    An S file illustrates some of the critical ideas.
Assignments.
  1. Assignment 1.
  2. Assignment 2.
  3. Assignment 3.
  4. Assignment 4.
  5. Exam data .