A simple baseline for travel time estimation using large-scale trip data

Abstract

The increased availability of large-scale trajectory data provides rich information for the study of urban dynamics. For example, New York City Taxi & Limousine Commission regularly releases source/destination information of taxi trips, where 173 million taxi trips released for Year 2013 [1]. Such a big dataset provides us potential new perspectives to address the traditional traffic problems. In this paper, we study the travel time estimation problem. Instead of following the traditional route-based travel time estimation, we propose to simply use a large amount of taxi trips without using the intermediate trajectory points to estimate the travel time between source and destination. Our experiments show very promising results. The proposed big data-driven approach significantly outperforms both state-of-the-art route-based method and online map services. Our study indicates that novel simple approaches could be empowered by the big data and these approaches could serve as new baselines for some traditional computational problems.

Learning Travel Time Distributions with Deep Generative Model

“…Our observations suggest that traffic conditions in the same time slot are not invariant across the temporal dimension. The assumption made in previous studies does not hold [27,38,39,41,42].…”

“…To address the weaknesses of road segment-based methods, route-based methods attempt to estimate the travel time of a route as a whole. There are two main threads for route-based methods, namely, nearest neighbors search [36,39] and trajectory regression [20,41,48]. Nearest neighbors search estimates travel time of a route by averaging travel times of the historical trajectories that have the closed origin and destination with the query route.…”

“…Many online web apps (e.g., Google Map and Uber) provide such service. Due to its great importance, significant research efforts have been made towards the accurate estimation of travel times [20,38,39,41]. However, existing studies only focus on the estimation of expected travel time.…”

“…Roads in central business district are frequently visited, while roads in rural or remote areas are rarely covered. It is reported that about 10% of paths cannot be estimated due to lack of neighbor trajectories [39]. Second, travel time heavily depends on real-time traffic conditions which are dynamic.…”

“…Second, travel time heavily depends on real-time traffic conditions which are dynamic. Almost all existing methods [20,25,38,39,42] assume that traffic conditions in the same time slot (e.g., 8:00am-9:00am on every Saturday) are temporally-invariant, when estimating travel times. The assumption does not hold because traffic conditions can be influenced by many random variables such as the daily activities of people, road works, and accidents.…”