UniTE—The Best of Both Worlds: Unifying Function-fitting and Aggregation-based Approaches to Travel Time and Travel Speed Estimation

Travel time and speed estimation are part of many intelligent transportation applications. Existing approaches rely on either function fitting or data aggregation represent different tradeoffs between generalizability accuracy. Function-fitting learn functions that map feature vectors of, e.g., routes to travel estimates, which enables generalization unseen routes. However, mapping imperfect offer poor accuracy in practice. Aggregation-based instead form estimates by aggregating historical data, traversal for This very high given sufficient data. they simplistic heuristics when insufficient is available, yielding generalizability. We present a Unifying approach Estimation (UniTE) combines function-fitting aggregation-based into unified framework aims achieve the available. demonstrate empirically an instance UniTE can improve accuracies 40–64% 3–23%, respectively, compared using only aggregation.