Google Transit Research Articles

Measuring Public-Transport Network Connectivity Using Google Transit with Comparison across Cities

The aim of the study was to construct a framework to determine interconnectivity among public transport routes using the information provided by Google Transit and to apply this framework to appraise and compare the network connectivity of Auckland, London, and Paris. Google Transit provides both spatial and network data that are sourced directly from transport agencies, thus making it an efficient tool for retrieving the data required to measure connectivity. This study contributes to previously-developed methodologies for determining connectivity by (a) including the qualitative measures, which are smoothness of transfer and information availability, along with the quantitative measures, and (b) using Google Transit as an alternative data source. The results showed that the overall public transport connectivity of the network in Paris is better than that in London and Auckland. Auckland’s network had the most poor connectivity values. Findings suggest that Auckland’s network would benefit from more integrated services.

En-Route Planning of Multi-Destination Public-Transport Trips Using Smartphones

Public transport travelers require well-integrated, real-time information systems to use a network. The objective of this study was twofold: (1) to develop a model to determine the effect of personalized information provision through smartphones on user ability to plan multi-destination trips, and (2) to understand user perception of riding public transport aided by real-time, multi-destination trip-planning smartphone applications. Auckland Pioneer was developed for multi-destination trip planning and integrates real-time public transport services with search functions for places of interest. A test trial was conducted in Auckland, New Zealand, and the findings show the effects of the application on multi-destination trip planning in comparison to traditional information sources such as Google Transit. This research contributes to existing literature by demonstrating the importance of personalized information en-route for user multi-destination trip planning. With information to assist trip planning and provide step-by-step guidance en-route, users tend to feel more confident to ride public transport.

Critical Appraisal of Web-Based Passenger Information Systems

Passenger information is vital for developing a user-friendly public transportation system. Websites are rapidly gaining popularity for public transport information dissemination, particularly due to their anytime-anywhere availability and their suitability for the multimodal applications and multilingual interface. Internet-based Passenger Information Systems (PIS), therefore, have become common in developed countries. The development of PIS for urban transport in India however, is at an experimental stage with very few operational deployments. This paper attempts to examine the current state-of-the-art features in Web-based passenger information systems in India and abroad, while critically evaluating the existing sources of public transport information in Ahmedabad as a case study. Ahmedabad, like several other Indian cities, has fixed-route regular bus services in conjunction with a recently-introduced bus rapid transit (BRT) system. The study compares the information content in printed transit timetables, Google Transit, and websites of the transit agencies, with the spatial dataset of public transport network prepared by integrating information from several sources. The results highlight the issues pertaining to accuracy, coverage, and timeliness of information contents available in developing countries, requiring innovative technological interventions to meet the growing information needs of commuters.

Energy Cost Minimization Based on Decentralized Reinforcement Learning With Feedback for Stable Operation of Wireless Charging Electric Tram Network

The convenience and energy efficiency of electric trams have resulted in their rapid adoption in many areas of the world. Despite trams’ many advantages, overhead lines installed along the track cause severe problems, such as safety issues, maintenance problems, and difficulty in renovating the urban road system. To solve these problems, a wireless charging electric tram system has been developed to replace the overhead lines, catenary, and pantograph. With wireless charging segments installed underground, a tram can operate freely without any support from overhead lines. In order to successfully commercialize a wireless charging tram system, it is essential to minimize the energy cost of the wireless charging tram system. In this article, a model of a wireless charging electric tram system and an energy cost minimization algorithm are proposed. The Manchester Metrolink tram network is modeled using the Google Maps API, the Google Transit API, and the Metrolink API to simulate the wireless charging system under massive public transportation network. To determine the minimum energy cost, an energy cost minimization algorithm for the wireless charging tram system is proposed based on decentralized multiagent reinforcement learning with feedback. The numerical results show that the proposed algorithm can reduce costs while ensuring stable operation.

Abstract C06: Travel burden to breast biopsy facilities by race and disadvantage in Chicago

Abstract Purpose: The purpose of this analysis is to explore travel burden to biopsy facilities incurred by breast cancer patients. Methods: We geocoded patient residence and biopsy facility addresses from Breast Cancer Care in Chicago, a population-based study of breast cancer patients diagnosed from 2005-2008 (n=989). Street network based drive times from patient residence to the facility where the patient received biopsy services were calculated using ArcGIS. Public transportation times were obtained using Google Transit ‘best route’ and set to arrive at the facility by 11:00am on a Thursday. The percentage of households owning a vehicle for each patient's census tract (US Census 2010) was used to calculate a weighted travel time from both driving and public transportation times. An additional ‘assigned’ travel time was set as the drive time when tract-level vehicle ownership was ≥50% and set as public transportation time when tract-level vehicle ownership was <50%. Kruskal Wallis and Wilcoxon Rank Sum tests were used to examine the differences between travel times by census tract level disadvantage (<1 SD below the mean, within 1 SD of the mean, >1 SD above the mean) and affluence (<1 SD below the mean, within 1 SD of the mean, >1 SD above the mean), race/ethnicity (nH black, Hispanic, nH white), highest education received (<high school, high school, >high school), annual household income (<30k, 30-75k, >75k), insurance status (none, public, private) and region of the city (south side vs. north side). Results: Patient residence and biopsy facility locations were available for 890 patients. Average weighted travel time was 17 minutes (range 0.5 to 85) and average assigned travel time was 11 minutes (range 0.3 to 74), whereas average drive time alone was 9 minutes (range 0.3 to 58) and average public transportation time alone was 45 minutes (range 3 to 190). Non-Hispanic (nH) black and Hispanic women had longer travel times than nH white women for all types of travel time (all p<0.01). Women living in census tracts of higher disadvantage and lower affluence also had longer travel times than their counterparts for each type of travel time (all p<0.001). Women with private insurance had shorter travel times that women with public insurance or women without insurance using the weighted travel time (p=0.01). Women living on the south side of Chicago had longer travel times that women living on the north side for all types of travel time (all p<0.0001). Conclusions: Longer travel times to biopsy services could potentially serve as a barrier causing delays in diagnoses. Prior analysis among this study population found no meaningful differences in travel time to breast cancer care facilities using drive time alone. The addition of public transportation times provides a more realistic estimate of travel burden among women in Chicago. We found that women of greater disadvantage had longer travel times to their biopsy facilities. We also found longer travel times among women living on the south side of Chicago, which is where the most predominately nH black and socioeconomically disadvantaged neighborhoods are located. Greater travel burden incurred by disadvantaged women may be contributing to Chicago's racial disparity in breast cancer stage at diagnosis. Future analyses will explore how travel burden to breast cancer care facilities relates to breast cancer outcomes. Citation Format: Jenna A. Khan, Garth Rauscher. Travel burden to breast biopsy facilities by race and disadvantage in Chicago. [abstract]. In: Proceedings of the Sixth AACR Conference: The Science of Cancer Health Disparities; Dec 6–9, 2013; Atlanta, GA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2014;23(11 Suppl):Abstract nr C06. doi:10.1158/1538-7755.DISP13-C06

Simplified model of local transit services

This paper discusses the development of a simplified model to efficiently represent local public transportation services in a large-scale travel demand model. The California Statewide Travel Demand Model (CSTDM) is a comprehensive model system designed and developed for use in transportation policy analysis and travel demand forecasting, including representation of both long and short distance transportation covering the entire state of California. A novel hybrid system is used to represent the full range of rail and bus transit services that are available. Rail services – including all long-distance rail, commuter rail and light rail services – are represented in the standard manner, using explicit node and link networks; the relevant in-vehicle and out-ofvehicle service characteristics for journeys are determined as standard skims of these networks. On-street bus services are not represented using explicit networks; rather, the relevant in-vehicle and out-of-vehicle service characteristics are determined using functions of other transportation network variables, land use descriptors and relevant policy indicators. These functions are simplified econometric models estimated using observations of transit service obtained from Google Transit Data Feeds. The network and simplified components are integrated in order to allow transit paths with both rail and on-street bus components to be considered by the various travel choice models included in the modelling framework. This hybrid system provides a suitable representation of transit for an area of such size. This facilitates consideration of transit service policies, while obviating the need for extensive transit coding, a daunting task for a large area.

Assessing public transport systems connectivity based on Google Transit data

A PT system consists of various physical features such as roads, railways, routes, and stops which are represented by a complex network of spatial and temporal data. Since these networks are usually very large and include millions of entities, it is difficult to assess PT systems. Assessment in this context is defined as the ability to extract and analyze data in an automated and recurring process so as to enhance decision making and to make it possible to compare between PT networks over time. The unified methodology that this work presents for extracting, storing and analyzing PT data enables relatively easy spatial analysis with GIS techniques based solely on: (a) Google Transit feeds and (b) Transportation networks. In order to implement this new methodology for analyzing a PT system, five connectivity indicators are introduced: (a) transportation network coverage level; (b) average speed; (c) intersection coverage level; (d) stop transfer potential; and (e) route overlap. This work demonstrates the proposed methodology by analyzing PT systems in Auckland (New Zealand), Vancouver (Canada), and Portland (Oregon, USA).

Hyperpaths in Network Based on Transit Schedules

The concept of a hyperpath was introduced for handling passenger strategies in route choice behavior for public transit, especially in a frequency-based transit service environment. This model for handling route choice behavior has been widely used for planning transit services, and hyperpaths are now applied in areas beyond public transit. A hyperpath representing more specific passenger behaviors on a network based on transit schedules is proposed. A link-based time-expanded (LBTE) network for transit schedules is introduced; in the network each link represents a scheduled vehicle trip (or trip segment) with departure time and travel time (or arrival time) between two consecutive stops. The proposed LBTE network reduces the effort to build a network based on transit schedules because the network is expanded with scheduled links. A link-based representation of a hypergraph with existing hyperpath model properties that is directly integrated with the LBTE network is also proposed. Transit passenger behavior was incorporated for transfers in the link-based hyperpath. The efficiency of the proposed hyperpath model was demonstrated. The proposed models were applied on a test network and a real transit network represented by the general specification of Google's transit feed.