Effects of Coverage Area Treatment, Spatial Analysis Unit, and Regression Model on the Results of Station-Level Demand Modeling of Urban Rail Transit (original) (raw)
Related papers
Regression Models for Predicting Rail Transit Ridership at the Station Level
2019
Methods for predicting ridership for future urban rail systems or extensions often have poor accuracy. One study shows that predicted ridership is overestimated by about 50%, on average, for a broad sample of urban rail systems worldwide. The ridership estimates produced by most transit agencies in the United States are not based on regression models. This thesis presents a framework for feature generation and regression modeling for estimating urban rail ridership in the United States. Features are generated using publicly available data from the US Census Bureau at the zip code level. Monte Carlo geographic sampling from zip code shapefiles generates features for each station on a rail network, representing characteristics within walking distance of that station. Network connections and travel times are used to generate a second set of features representing characteristics within commuting distance of each station. Several models are developed using different regression types and are compared in terms of accuracy and selected features. Some of the generated models provide system-level ridership predictions within 20% of the true value for a sample set of six US urban rail systems.
Applied Spatial Analysis and Policy
Boarding and alighting modeling at the bus stop level is an important tool for operational planning of public transport systems, in addition to contributing to transit-oriented development. The interest variables, in this case, present two particularities that strongly influence the performance of proposed estimates: they demonstrate spatial dependence and are count data. Moreover, in most cases, these data are not easy to collect. Thus, the present study proposes a comparison of approaches for transit ridership modeling at the bus stop level, applying linear, Poisson, Geographically Weighted and Geographically Weighted Poisson (GWPR) regressions, as well as Universal Kriging (UK), to the boarding and alighting data along a bus line in the city of São Paulo, Brazil. The results from goodness-of-fit measures confirmed the assumption that adding asymmetry and spatial autocorrelation, isolated and together, to the transportation demand modeling, contributes to a gradual improvement in ...
Factors that Influence Urban Streetcar Ridership in the United States
Transportation Research Record, 2013
meaningful ridership estimates for new streetcar systems or extensions with the use of travel demand models only. A direct ridership model is one that has its basis in regression analysis, which measures empirical relationships through statistical analysis of station ridership and local station characteristics. Such models are directly and quantitatively responsive to land use and transit service characteristics within the immediate vicinity and within the catchment area of transit stations. They are used to predict ridership at individual stations on the basis of local station area and system characteristics. Although streetcar systems sometimes are used for traditional commute trips, experience with transit agencies suggests that they predominantly provide access and circulation for downtown or city center areas. Streetcar systems commonly serve tourists, and they often duplicate existing transit service provided by bus (3). Thus it was expected that individual station-area characteristics greatly affected boarding and overall ridership projections. Limited research on urban streetcar ridership factors, along with the challenges to assess the streetcar market with regional travel demand models, highlight the value of the effort presented here. The findings are intended to help agencies and practitioners to develop a more accurate assessment of the relative benefits of streetcar lines. This research built on past transit ridership forecasting research, which focused primarily on light rail, commuter rail, and bus systems. With the recognition that variables that affect streetcar ridership may differ from those for other transit systems, this study analyzed ridership and characteristics for existing streetcar systems in Portland, Oregon, and Seattle and Tacoma, Washington. A directional, stationlevel, ordinary least squares (OLS) regression model approach was selected to predict streetcar boardings. The sections of the paper that follow describe the research. They include a literature review, data collection, methodology, model results and analysis, conclusions, and further research. Literature review Direct ridership forecasting is a growing field of research. Numerous studies have been conducted to estimate the factors that influence ridership for various transit systems (4-12). Table 1 provides an overview of research conducted on direct ridership forecasting. Most of the research references earlier work on the relationship between built environment, "D" variables (e.g., density, diversity, design), and travel demand (13-15) and the attempt to incorporate these variables into the ridership models. Some form of population or employment variable, or both, was included in each of the models studied. Typically, the variable represented the population, household, or employment density within a certain area around the given station. Station parking also was found to influence ridership for
Identifying the Factors that Influence Urban Public Transit Demand
ArXiv, 2021
The rise in urbanization throughout the United States (US) in recent years has required urban planners and transportation engineers to have greater consideration for the transportation services available to residents of a metropolitan region. This compels transportation authorities to provide better and more reliable modes of public transit through improved technologies and increased service quality. These improvements can be achieved by identifying and understanding the factors that influence urban public transit demand. Common factors that can influence urban public transit demand can be internal and/or external factors. Internal factors include policy measures such as transit fares, service headways, and travel times. External factors can include geographic, socioeconomic, and highway facility characteristics. There is inherent simultaneity between transit supply and demand, thus a two-stage least squares (2SLS) regression modeling procedure should be conducted to forecast urban ...
Sustainability
This study investigates changes in population density in 454 block groups within a one-mile buffer around rail transit stations (the study area) in the Dallas-Fort Worth (DFW) metropolitan area. The research uses three analysis approaches to explore a correlation between proximity to rail stations and population density changes. Changes in population density between 2000 and 2014 are calculated. Changes in population density in the study area are compared to the remainder of the block groups within the four counties served by the same rail transit systems. An innovative approach is employed to select the best regression model using the data of the study area. A relationship between the independent variables and the changes in population density is formulated. The proximity of block groups in the study area to the nearby highway ramps or city centers is also investigated during the study period. Results show that it has a positive impact on population density. Changes in population d...
How to Increase Rail Ridership in Maryland? Direct Ridership Models (DRM) for Policy Guidance
The state of Maryland aims to double its transit ridership by the end of 2020. The Maryland 5 Statewide Transportation Model (MSTM) has been used to analyze different policy options at 6 a system-wide level. Direct ridership models (DRM) estimate ridership as a function of 7 station environment and transit service features rather than using mode-choice results from 8 large-scale traditional models. They have been particularly favored for estimating the benefits 9 of smart growth policies such as Transit Oriented Development (TOD) on transit ridership 10 and can be used as complementary to the traditional four-step models for analyzing smart 11 growth scenarios at a local level and can provide valuable information that a system level 12 analysis cannot provide. In this study, we developed DRMs of rail transit stations, namely 13 light rail, commuter rail, Baltimore metro, and Washington D.C. metro for the state of 14 Maryland. Data for 117 rail stations were gathered from a variety of sources and categorized 15 by transit service characteristics, station built environment features and social-demographic 16 variables. The results suggest that impacts of built environment show differences for light rail 17 and commuter rail. For light rail stations, employment at half-mile buffer areas, service level, 18 feeder bus connectivity, station distance to the CBD, distance to the nearest station, and 19 terminal stations are significant factors affecting ridership. For commuter rail stations only 20 feeder bus connection is found to be significant. The policy implications of the results are 21 discussed.
Transportation Research Record, 1991
Historical data on a small numb r of econom ic demographic, and tran pol'lation v, riable from 1958 to 1986 were an;ilyzcd by multiple regressi n technique to develop two models for forecasting tran~it ridership in Honolulu. A model predicting revenue trips and another for linked trips were consistent in their determination that the same five variables could account for 97 to 98 percent of the variance in bus ridership over this 29-year period. The four major variables were per capita income, employment, fares, and size of bus fleet, with a dummy variable included for strikes. The income elasticity for transit demand was found to be negative , indicating that mass tran it is an inferior good. The model foreca ts a continuing declin • in bu • ride1 hip for Honolulu , mainly caused by this effect. The foreca ting model$ for rapid transit rider hip forH nolulu are examined, and al.ternative approaches to assessing demand elasticities are discussed. The advantages of using aggregate historical data and regression analyses for developing inexpensive forecasting models from time series data are emphasized. Two multivariate models to forecast transit ridership for Honolulu using aggregate variables are presented and discussed with respect to different modeling approaches and thei1 applications. The two models use the statistical technique of multiple regression that is widely used in economic forecasting and model construction in the other social sciences (1,2). This approach is most commonly used in transportation to study trends in time series data (3-7) and it is particularly useful for analyzing secondary sources of historical data (8,9). As such, it is well suited for long-range planning and it can be a valuable tool for transportation planners who have only limited resources available to them. ELASTICITY OF TRANSIT DEMAND The demand for transit (transit ridership), like that for any product, is related to two variables: price and income. The price relationship is best known. The demand for a product
How to Increase Rail Ridership in Maryland: Direct Ridership Models for Policy Guidance
Journal of Urban Planning and Development, 2016
The state of Maryland aims to double its transit ridership by the end of 2020. The Maryland Statewide Transportation Model (MSTM) has been used to analyze different policy options at a system-wide level. Direct ridership models (DRM) estimate ridership as a function of station environment and transit service features rather than using mode-choice results from large-scale traditional models. They have been particularly favored for estimating the benefits of smart growth policies such as Transit Oriented Development (TOD) on transit ridership and can be used as complementary to the traditional four-step models for analyzing smart growth scenarios at a local level and can provide valuable information that a system level analysis cannot provide. In this study, we developed DRMs of rail transit stations, namely light rail, commuter rail, Baltimore metro, and Washington D.C. metro for the state of Maryland. Data for 117 rail stations were gathered from a variety of sources and categorized by transit service characteristics, station built environment features and social-demographic variables. The results suggest that impacts of built environment show differences for light rail and commuter rail. For light rail stations, employment at half-mile buffer areas, service level, feeder bus connectivity, station distance to the CBD, distance to the nearest station, and terminal stations are significant factors affecting ridership. For commuter rail stations only feeder bus connection is found to be significant. The policy implications of the results are discussed.
2012
Geographically weighted regression (GWR) and Multiple Linear Regression (MLR) models have been applied to derive the spatial structure of urban heat island (UHI) in Wroc aw, SW Poland and compared. It was found that GWR is better suited for spatial modeling of UHI than MLR, as it takes into account non-stationarity of the spatial process. Both local and global models were extended by the interpolation of regression residuals, and used for spatial interpolation of the UHI structure. The combined: GWR + interpolated regression residuals (GWRK) approach is recommended for spatial modeling of UHI.