History’s Principal Components Repeat Themselves: Using Data for Traffic Predictive Control Samuel Coogan University of California, Los Angeles (UCLA) The advent of ubiquitous traffic sensing provides unprecedented real-time, high-resolution data of traffic conditions that elucidate historical trends and current traffic conditions, yet traditional signal control approaches are designed to operate with relatively limited or no real-time and/or historical data. In this talk, we propose using principle component-based decomposition techniques to learn trends in historical data that are then used to make real-time predictions of traffic flow minutes or hours into the future. We show that such predictions lead to practical, preemptive control schemes that accommodate the predicted traffic conditions rather than average traffic conditions, thus improving intersection performance and efficiency.