NSW-EPNews: News-Aligned Electricity Price Forecasting Benchmark

1.2 GB benchmark dataset pairing high-frequency electricity price data from New South Wales with temporally aligned market news. It was created by Zhaoge Bi and last updated on May 7, 2026, to evaluate the use of textual context in forecasting models. The dataset demonstrates that current numerical models and large language models do not yet use market news consistently or reliably.

Use Cases

• Benchmarking numerical forecasting models based on high-frequency electricity price data.
• Evaluating large language models' ability to incorporate textual market news for price prediction.
• Developing multimodal models that fuse time-series and text data for energy market analysis.
• Studying the correlation between specific news events and electricity price volatility in New South Wales.

Strengths

• Dataset size is 1.2 GB, indicating a substantial collection of paired data.
• Explicitly designed as a benchmark for evaluating model performance on a specific task.
• Released under a permissive CC-BY-4.0 license, facilitating reuse and sharing.

Limitations

• Row count and specific column definitions are unknown, limiting suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Geographic coverage is limited to New South Wales, Australia.

Provenance

Source

Author Zhaoge Bi, hosted on figshare.

Collection Method

Likely involves collecting and aligning public electricity market data with contemporaneous news reports.

Freshness

Last updated 2026-05-07 11:03:07

Geography

New South Wales, Australia