Thanks to Drillinginfo’s PRT next-day load forecasts, utilities, power marketers, and other PRT users were prepared for the frigid temperatures that swept through Texas a few weeks ago. The cold snap provided an opportunity to demonstrate how machine learning technology can deliver accurate, actionable information. When it comes to load forecasting, accuracy matters.
PRT’s next-day load forecasts vastly outperformed the Electric Reliability Council of Texas (ERCOT) Independent System Operators’ (ISO) own forecasts.
Rob Allerman, Director of Analytics at Drillinginfo, first uncovered a bullish load trend published by ERCOT on Nov. 7 and called out ERCOT’s over-projection for the cold snap on Monday, Nov. 12, in the daily reports. Not only did this trend continue, but the bullish sentiment intensified into the week. PRT’s adaptive machine learning stayed on track while Allerman’s watchful eye kept PRT users informed as Texas entered its first cold snap of the season.
From Nov. 12 to Nov. 15, PRT’s peak mean absolute percentage error (MAPE) was less than 2 percent, while ERCOT came in more than two percentage points higher at 4.26 percent.
Latest posts by Enverus (see all)
- Do Renewable Asset M&A Multiples Point to a Bubble? - June 14, 2021
- Hellman & Friedman Completes Acquisition of Enverus - June 10, 2021
- Oil Sands Forecasting — More Accurate Than Your Local Meteorologist - June 7, 2021