The California ISO Department of Market Monitoring (DMM) has posted the Review of the Mosaic Quantile Regression for estimating net load uncertainty, and has scheduled a stakeholder web conference on Dec. 1, 2023 to discuss reported analysis and findings.
In February, the California ISO implemented a new calculation of real-time net load uncertainty used in the flexible ramping product demand and in the Western Energy Imbalance Market resource sufficiency evaluation. This new calculation, the mosaic quantile regression, uses historic data to estimate the relationship between extreme outcomes of net load uncertainty, and load, solar, and wind forecasts.
DMM replicated the ISO methodology and conducted statistical analysis on results. Key findings of this analysis include the following:
- The mosaic quantile regression model has limited predictive capability for forecasting net load uncertainty.
- The results from the mosaic quantile regression closely resemble the histogram model, highlighting the weak relationship between net load uncertainty and the mosaic variable, the selected predictor for net load uncertainty.
- The quantile regression relies on a limited sample size, which may compromise the validity of the regression model.
- The performance of the quantile regression and histogram methods is quite similar, with neither method showing clear dominance in accuracy or efficiency.
- The mosaic quantile regression model has a specification error involving the use of unintended information in forecasting, along with unnecessary steps that do not influence the outcome.
Information related to market monitoring is available on the Market Monitoring webpage.
Meeting details
Date: Dec. 1, 2023
Time: 10 a.m. - 11 a.m. Pacific Time
Add to calendar
Web and audio participation information
Please pre-register at https://ems8.intellor.com?do=register&t=1&p=849501 to receive a join link for the event.
Technical support: call Webex Support at 888-793-6118
