Course Description

Governments forecast economic indicators (e.g., GDP, job growth, etc.); businesses forecast sales; portfolio managers forecast asset return; the list goes on. Accurate forecasts are critical to robust organizational decision-making. This course (CMU course number 73-423) will introduce students to modern methods for forecasting in economic and business applications. Topics covered include Bayesian, statistical, and online learning approaches to forecast construction and assessment, univariate and multivariate time series models and algorithms, and principled combination of multiple methods and data sources along with subject matter expertise to improve performance. Methods will be motivated by applications in macroeconomics, technology, marketing, and finance, with cases drawn from forecasting processes in a variety of business and government organizations. Students will implement forecasting methods in R, including in a real data forecasting competition. The primary external text for the course is Forecasting: Principles and Practice, by Rob Hyndman and George Athanasopoulos, with substantial content aggregated from other sources.

Course Materials

The following files, derived from the lecture slides for the course and containing both text and R code, are provided as-is, as a resource for students and researchers interested in the topics. They are currently in draft form and are likely to contain errors both big and small. If you have questions, comments, suggestions, or criticisms of the material, please contact me. Additional course materials, including syllabi, problem sets, practice problems, and project assignments, may be available upon request.

  1. Introduction
  2. Loading and Visualizing Time Series Data in R (Kaggle link)
  3. Methods and Motivation
  4. Evaluating Forecasting Methods
  5. The Statistical Approach
  6. Empirical Risk Minimization
  7. Applying Empirical Risk Minimization
  8. Multivariate Forecasts
  9. Regularization
  10. Uncertainty Quantification
  11. Bayes
  12. Applying Bayesian Methods
  13. Additive Component Models
  14. Autoregression Models
  15. ARIMA
  16. State Space Models
  17. Online Learning and Regret Minimization
  18. Online Learning - Algorithms
  19. Model Combination
  20. Factor Models
  21. Machine Learning
  22. Neural Networks
  23. Judgment