Electric Load Forecasting II: Advanced Topics and Case Studies

The standard course outline is listed below. The course is being customized based on the background and needs of attendees. See THIS BLOG POST for my teaching method

Course Outline

Out of sample tests

  • Error analysis

  • Cross validation

  • Sliding simulation

Weather station selection

  • Two fundamental questions

  • A common method

  • Unconstrained weather station selection

  • Seven-step implementation

Outlier detection and data cleansing

  • Definitions of outlier

  • Three examples

  • Hidden outlier

  • A modeling approach to outlier detection and data cleansing

More about recency effect

  • How many lagged temperatures can we afford

  • “Optimal” combination of lagged and average temperatures

  • Recency effect in hierarchical load forecasting

  • Search algorithms for recency effect modeling

Combining forecasts

  • Motivation

  • Forecast combination methods

  • Practical considerations

Case studies (computer lab session)

  • Cross validation

  • Weather station selection

  • Outlier detection and data cleansing

  • Recency effect modeling

  • Forecast combination

Emerging topics (optional)

  • ARIMA models for electric load forecasting

  • Grouping and clustering methods

  • Hierarchical load forecasting

  • Probabilistic electric load forecasting

  • Retail energy forecasting

  • Other emerging topics