Example: ERCOT Energy vs Dallas Weather

A real research investigation run on Lomita, demonstrating data discovery, agent chain execution, and statistical analysis.

The question

"Does ERCOT energy demand correlate with Dallas weather extremes?"

What happened

Time: ~5 minutes from question to completed report

Discovery Agent (30 seconds)

Searched the catalog and found:

  • ercot-public-api — ERCOT locational marginal pricing and grid load data
  • noaa-cdo-weather — Historical temperature and weather for Dallas
  • eia-electricity-generation — Real-time electricity generation by fuel type
  • eia-steo-forecasts — Short-term energy outlook

Integration Engineer (2 minutes)

Built data pipelines for each source. Three Integration Engineers ran in parallel — each building a separate pipeline.

Quant Analyst (2 minutes)

Ran statistical analysis:

  • Pearson correlation between temperature and ERCOT load
  • Granger causality tests (does temperature predict demand?)
  • Lag analysis (how many hours of lead time?)
  • Regime analysis (does the relationship differ in summer vs winter?)

Research Narrator (1 minute)

Compiled findings into an executive report.

The verdict

SUPPORTED (Green)

Key findings:

  • Strong positive correlation between temperature extremes and ERCOT load (r = 0.67)
  • Bidirectional Granger causality (p < 0.01) — temperature predicts demand
  • Strongest effect at 1-hour lag in summer (AC load)
  • Winter heating demand also significant but weaker

What the graph shows

After this research completed, the knowledge graph showed:

  • Hypothesis node (green) connected to 4 source nodes
  • Entity nodes: "Dallas Temperature", "ERCOT Load", "ERCOT LZ_NORTH Price"
  • CORRELATES_WITH edges with r-values between entities
  • Cross-domain connections (weather → energy) visible in the graph layout

Delivery

The report was auto-delivered to email and manually delivered to Slack — both via the chat:

Deliver this report to [email protected]

Try it yourself

This exact research can be reproduced on any Lomita instance. Go to the Explore page and type:

Research whether ERCOT energy demand correlates with Dallas weather extremes