Researchers have introduced a new model called RiskTraf for predicting traffic flow. Unlike existing models that often ignore or misrepresent raw measurements of speed and occupancy, RiskTraf uses all three variables - flow, speed, and occupancy - to make more accurate predictions. The model works by learning a residual head from historical speed and occupancy data, which helps it adapt to different traffic conditions without modifying the underlying backbone network. The authors claim that RiskTraf outperforms other methods in extensive experiments. --- Why it matters: This matters because accurately predicting traffic flow is crucial for optimizing traffic management systems and reducing congestion. By using a more comprehensive set of variables and adapting to changing traffic conditions, RiskTraf has the potential to improve the accuracy of traffic predictions and inform more effective decision-making. Source: https://arxiv.org/abs/2608.20656