Many forecasting applications and decision making processes require explanatory variables, making regression analysis an attractive solution in the business world. This course offers a holistic view of a regression-based approach to forecasting. We will discuss several representative regression techniques for forecasting both continuous variables and categorical variables, with the output in both point and probabilistic forms. We will also discuss several important concepts such as model selection and regularization. Real-world data and examples will be used to illustrate the theory. The course also includes about two hours of lab session for the attendees to practice the regression techniques in SAS.
Course Outline
Forecasting problems
Error analysis
Micro and macro views of model selection
Linear regression
Model selection methods
Lab session I
Ridge regression and lasso
Support vector regression
Robust regression
Quantile regression
Fuzzy regression
Logistic regression
Lab session II