Stat 350 syllabus

Applied Time Series Analysis

 

Bulletin Description Fundamentals concepts; classical regression models as

forecasting models, exponential smoothings, stationary and nonstationary

models, additive and multiplicative decompositions, moving average,

autoregressive, ARMA and ARIMA processes, estimation in MA, AR, ARMA and

ARIMA processes. Box-Jenkins methodology, computer aided modeling,

applications.

Prerequisite ST 310 or ST 315 or ST 320 or ST 335. Computer Lab fee.

Text Time Series Forecasting: Unified Concepts and Computer Implementation,

by Bruce L. Bowerman and Richard T. O'Connell, 3rd edition, Duxbury Press, 1991.

 

Coverage Material to be selected by the instructor

 

Learning Objectives This course is designed to give applied yet sophisticated

presentation of classical and modern statistical techniques that are useful in short

term forecasting of time series data. The development of topics begins with the

basic simple linear regression and moves into complex regression based

prediction using dummy variables and smoothing methods. Additive and

multiplicative decompositions, based on four major aspects of a time series data

are discussed and finally the modern technique, known as Box-Jenkins

methodology is introduced where the nature of time series is identified using the

concept of auto correlation and partial auto correlation. Estimation of parameters

is discussed. Statistical computer software is intended to enhance the facility with

applications of various techniques covered in this course.

 

Last Updated February 18, 2014