統計科学講演会（2025年5月27日開催）「Functional Regression with Derivatives for Traffic Prediction」のご案内

講演概要：

　To predict traffic congestion within a short time frame, we propose a generalized functional linear regression model using traffic speed trajectories and their first two derivatives as functional predictors.
As the derivatives are not directly observed, they are estimated and represented, along with the trajectories, via the Karhunen–Loève expansion. The model captures both integrated and individual predictor effects, with weight parameters indicating their relative importance.
An application to freeway data shows that incorporating trajectory derivatives improves prediction accuracy.