California Renewable Curtailments Study

Foreword

The work done in this project was done in an academic context as an independent study. The original course proposal set out to define a domain-oriented problem to learn and apply statistical learning methods. Many thanks go out to Peter Alstone for his guidance, metorship and support.

The original course goals focused on developing a familiarity with the opensource ML and statistical learning ecosystem.

Table of Contents

Contents

• Motivation
  • Background
  • Problem Statement
• Data Sources
  • Curtailments Data
  • Historic Weather Forecasts
  • CA County Boundaries
  • CEC Powerplant Locations
• Methods
  • Example: Predicting Daily Curtailment Events
  • Single Model Run
  • Multi-Model Run
  • Visualize Results
  • Statistical Learning Process
  • Tooling
• Summary of Prediction Results
• Conclusion
  • Next Steps
• References