June 16 Data Science Community – SMART-STLF: A Situational, Multi-Source, Adjustable, Reasoning Tool for Short-Term Load Forecasting Under Typical and Extreme Weather Conditions

Join the Data Science Community on Tuesday, June 16, 2026 at 1:00 PM CT for a presentation with Q&A/open discussion on, “SMART-STLF: A Situational, Multi-Source, Adjustable, Reasoning Tool for Short-Term Load Forecasting Under Typical and Extreme Weather Conditions” led by Gregory Labbe, Director, Corporate Analytics and Mehdi Ganjkhani, Ph.D., Senior Data Scientist, Analytics at The Energy Authority.

Session Description:

Short-term load forecasting (STLF) is a critical operational task for electric utilities, particularly during periods of weather-driven demand volatility.

This presentation introduces SMART-STLF, an interactive decision-support tool developed at The Energy Authority (TEA) to enhance forecaster situational awareness and enable real-time, data-informed adjustments to baseline load forecasts.
The tool is built around four core capabilities:

• Situational Awareness: Real-time NOAA weather alerts are surfaced directly to traders, providing early warning of conditions that may impact forecast accuracy.
• Multi-Source Forecasting: Side-by-side weather forecasts from NOAA and DTN—along with corresponding load implications—allow traders to evaluate uncertainty across sources and scenarios.
• Adjustable Forecasting: Two complementary approaches are supported:
  • Scenario-based forecasting: Traders can specify a target temperature scenario to generate a what-if load forecast, which is particularly valuable for stress-testing under extreme or hypothetical weather conditions.
  • Historical similarity adjustment: For a given target day, the tool identifies historical days with the most similar weather profiles, analyzes the gap between actual load and the baseline forecast on those days, and uses a machine learning model to recommend an adjustment to the current baseline forecast — along with a confidence interval.
• GenAI-Driven Reasoning: A large language model (LLM) provides real-time narrative context for each forecast day, including characterization of the weather regime (typical vs. extreme), an adjustment confidence assessment, and operational recommendations — translating quantitative outputs into actionable guidance.

SMART-STLF bridges the gap between data-driven forecasting models and operational decision-making, enabling more confident actions under both routine and high-stakes conditions.

*UAI Utility Members can register and access all nine of our Analytics Communities via our Community Page. Not a part of the Data Science Community and want to join this session? Please use our direct link here to request access. You will find the Teams meeting credentials to the upcoming meeting at the top of the feed on this page. If you have any questions or run into any issues, please contact us at info@utilityanalytics.com.