AI

When better traffic forecasts fail to improve signal control: a layered diagnostic study of forecast-to-decision value

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 08/10/2026, 02:18:24 UTC.

Briefings

A layered diagnostic study using 29 days of reconstructed demand in Xuancheng, China, found that improved traffic forecasts do not always improve signal control. While entry-level and movement-level forecasts reduced mean absolute error by 4.03% and 3.92% respectively, decision-time performance varied significantly. On frozen test dates, causal forecasts and an event oracle increased queue vehicle-seconds by 6.09% and 3.39% relative to matched no-future rollout, while the oracle reduced spillback exposure by 3.78%. The study highlights that forecast value depends on temporal observability, action identifiability, dynamics consistency, and objective alignment.

Key points

• Entry-level and movement-level forecasts reduced mean absolute error by 4.03% and 3.92% relative to historical means in Xuancheng, China.
• On frozen test dates, causal forecasts and an event oracle increased queue vehicle-seconds by 6.09% and 3.39% relative to matched no-future rollout, while the oracle reduced spillback exposure by 3.78%.

Context

This research investigates the gap between improved traffic forecasts and better signal-control decisions, using reconstructed demand data from Xuancheng, China, to evaluate various forecasting methods and their impact on intersection control.

Limitations

The study covers 29 days of reconstructed demand with seven dates reserved for testing. On frozen test dates, causal forecasts and an event oracle increased queue vehicle-seconds by 6.09% and 3.39% relative to matched no-future rollout, while the oracle reduced spillback exposure by 3.78%. Paired-day bootstrap intervals cross zero, indicating statistical uncertainty in the spillback reduction claim.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 04/11/2026, 02:07:39 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeidophj4u6i6vqpd2mbzhkjxpvydvgc73c6rxhxsmt47krv24udehu