520—Forecasting and Time Series. [=MGSC 520] (3) (Prereq: STAT 516 or MGSC 391) Time series analysis and forecasting using the multiple regression and Box-Jenkins approaches.
Sample Course Homepage: Recent Semester
Usually Offered: Alternating Fall Semesters and in MGSC
Purpose: To show the student how to recognize time series data, and to acquaint the student with the peculiarities of this kind of data. The appropriate questions to ask of the data and the general approaches which are particular to these data are examined in detail.
Current Textbook: Time Series Analysis with Applications in R, 2nd edition, by Cryer, J. and Chan, K., Springer, 2008.
|Introduction to time series. Autocorrelated data. Stationarity and trends||1 week|
|Autoregressive, moving average, and autoregressive moving average models. Random walk model. Seasonal models. The B-J notation. Integrated models.||2 weeks|
|The concept of identification. Autocorrelation and partial autocorrelation function. Behavior of sample estimates||2 weeks|
|Diagnostic checking. Residual analysis. Result of over and underspecifying the model.||2 weeks|
|Forecasting. Optimal forecasts and interval forecasts. Forecasting nonstationary models. Updating the forecasts. Exponential smoothing. Backcasting.||2 weeks|
|Details of analysis of several real socio-economic data sets.||2 weeks|
|The transfer function model and their identification. Cross covariance. Pre-whitening. Diagnostics. Forecasting.||2 weeks|
|Intervention analysis. Identification and estimation. The transfer function model.||1 week|
The above textbook and course outline should correspond to the most recent offering of the course by the Statistics Department. Please check the current course homepage or with the instructor for the course regulations, expectations, and operating procedures.
Contact Faculty: Joshua Tebbs