Typically offered during the spring semester.

Catalog description:

Introduction to quantitative prediction techniques using historical time series. Involves extensive use of interactive computing facilities in developing forecasting models and considers problems in design and updating of computerized forecasting systems. Cross-listed with STA 483. Credit not awarded for both this course and DSC 444.

Prerequisite:

Probability and Statistic STA 462 or equivalent.

Objectives:

• Apply the knowledge of statistics to the area of forecasting.
• Use regression, averaging, decomposition, smoothing, and ARIMA techniques for developing short term forecasting models.
• Develop time series models, causal models, and models that account for seasonality to make forecasts and estimate prediction intervals.
• Use Excel and SAS to support forecasting activities.

Required topics (approximate weeks allocated):

• Introduction to forecasting and use of spreadsheets (1)
  • types of methods
  • types of models
  • evaluation of models
• Regression analysis forecasting methods (3)
  • least squares parameter estimates
  • estimates of variance of error and parameters
  • prediction intervals for individual and cumulative forecasts
  • model evaluation
• Smoothing forecasting methods (6)
  • single and double moving averages
  • single and double exponential smoothing
  • direct smoothing
  • seasonal smoothing models
  • estimation of variance of error and parameters
  • prediction intervals
  • tracking signals and adaptive smoothing
• Autoregressive forecasting methods (4)
  • general model and special cases AR(p), MA(q), and ARMA(p,q)
  • interpretation of autocorrelations
  • parameter estimation
  • model evaluation
  • prediction intervals
• Exams/Review (1)