Mobility Vitality: Assessing Neighborhood Similarity Through Transportation Patterns In New York City

Full text

Mobility Vitality: Assessing Neighborhood
Similarity Through Transportation Patterns In
New York City
Dan Qiang #Ñ
Platial Analysis Lab, Department of Geography, McGill University, Montréal, Canada
Grant McKenzie #Ñ
Platial Analysis Lab, Department of Geography, McGill University, Montréal, Canada
Abstract
Though numerous studies have examined human mobility within an urban environment, few have
explored the concept of urban vitality purely through the lens of urban transportation. Given the
importance of different modes of transportation within a city, such analysis is necessary. In this
short paper, we introduce the novel concept of mobility vitality by integrating human mobility and
urban vitality, offering a multilayered framework to assess the degree of transportation and mobility
within and between regions. The mobility patterns of three transportation modes, namely subway,
taxicab, and bike-share, are first examined independently. These patterns are then aggregated to
form the composite measure of static mobility vitality. Through this measure, we evaluate similarities
between neighborhoods. Our results observed significant spatial differences in the travel patterns of
three transportation modes on weekdays and weekends. Moreover, neighborhoods with high static
mobility vitality have relatively similar mobility patterns. Ultimately, this approach aims to find
neighborhoods with imbalanced transportation infrastructure or inadequate public.
2012 ACM Subject Classification Information systems
→
Geographic information systems; Applied
computing →Transportation
Keywords and phrases mobility vitality, mobility similarity, transportation, bike-sharing, taxi,
subway, New York City
Digital Object Identifier 10.4230/LIPIcs.GIScience.2023.60
Category Short Paper
1 Introduction
In 1961, the urban activist Jane Jacobs introduced the concept of urban vitality as a qualitative
measure of a city’s pulse [
2
]. The idea suggests that varying tempos of human activities
and pedestrian flow can all be employed to differentiate regions. For decades, most of the
research related to this concept was done using qualitative surveys, demographic studies, and
narrative analysis. The difficulties with such approaches are costly and labor-intensive and
are prone to subjective biases. The recent dramatic growth of publicly accessible activity
and mobility data has set the stage for alternative approaches to assessing urban vitality.
Despite a large body of literature targetting the extraction of individual human mobility
patterns and their accompanying impact variables [
1
,
7
], little attention has been paid to
urban dynamics characterized purely by individual movement. Recently, a growing number
of research teams have focused on temporal characteristics of mobility to better understand
urban vitality [
3
]. For instance, Sulis et al. [
6
] examined smart-card rail trips to assess
spatiotemporal variation in urban vitality in London. They produced a set of three dynamic
properties, namely the number of people, the continuity, and the fluctuations of this presence
over particular intervals of time. Similarly, Zeng et al. [
9
] created a new index to measure
urban vitality based on records from a bicycle-sharing system. Further work has demonstrated
©Dan Qiang and Grant McKenzie;
licensed under Creative Commons License CC-BY 4.0
12th International Conference on Geographic Information Science (GIScience 2023).
Editors: Roger Beecham, Jed A. Long, Dianna Smith, Qunshan Zhao, and Sarah Wise; Article No.60; pp. 60:1–60:6
Leibniz International Proceedings in Informatics
Schloss Dagstuhl – Leibniz-Zentrum für Informatik, Dagstuhl Publishing, Germany

60:2 Mobility Vitality
that lively regions of a city correlate with taxi drop-off locations [
10
]. A variety of research
has shown that urban vitality/vibrancy can be measured through data ranging from social
check-ins and points of interest to trajectories and mobile phone data [4, 8].
Though progress is being made, research focused exclusively on mobility as a measure of
urban vitality is lacking [
11
]. In exploring the vitality of different parts of a city through a
mobility lens, one is able to identify the impact that access to different mobility modes, has
on city cohesion. Furthermore, a combination of mobility signatures can be used as a measure
through which different regions of a city can be compared [
5
]. Urban and transportation
planners can use such a measure to better understand the impacts of policy decisions on
the vibrancy and vitality of the city as a whole. Through integrating human mobility with
urban vitality, we proposed the novel concept of mobility vitality, serving as a multilayered
framework to evaluate the degree of transportation and mobility within a space. In this
preliminary work, we aim to address the following two research questions (RQ).
RQ1
Can a region, e.g., neighborhood, be quantified by the mobility patterns of different
modes of transportation that exist and traverse the region? Furthermore, do these
patterns vary by mode and region?
RQ2
Can mobility vitality, as represented by a combination of mobility patterns, be used
to compare and differentiate regions within the same city?
We address these questions through an analysis of three different modes of transportation
within New York City (NYC). As the most densely populated city in the United States,
NYC’s transportation ecosystem is both complex and extensive. The scale of our analysis is
neighborhoods within the five boroughs of NYC and the extent of analysis varies based on
the service area of each transportation system.
2 Data and Analysis
To start, three data sets representing three very different modes of transportation were
collected. These include bike-share, subway (rail), and taxicab data. We restricted our
analysis to May 2019, cleaned the data to remove errors, and aggregated the month of data
to days in a typical week. We use this week as a representative sample of transportation
usage in NYC. May was chosen due to the limited holidays, historically decent weather, and
fewer people on summer vacation. For micro-mobility, we accessed data for the widely used
bicycle-sharing system, Citi Bike
1
. Citi Bike is a privately operated docking station-based
bike-sharing system. Citi Bike trip data include the start and end times of each trip as well
as the origin and destination stations. For mid-sized transportation, we accessed trip data
for yellow taxis
2
. The yellow taxi trip records include fields capturing pick-up and drop-off
dates, times, and locations. For mass transit, we analyzed turnstile data of the NYC subway
system
3
. These data report an accumulated number of entrances and exits, per station at a
four-hour temporal resolution. All data were cleaned to remove erroneous trips (e.g., those
that were one minute in length, 200 miles, etc).
Next, we intersected the trip data with the NYC neighborhood boundaries
4
to assign trip
volume for each of the three services to each neighborhood in NYC. The assigned volume
includes both origins (entries) and destinations (exits). More specifically, the numbers of
1https://citibikenyc.com/system-data
2https://data.cityofnewyork.us/Transportation/2019-Yellow- Taxi-Trip-Data/2upf-qytp
3https://data.ny.gov/Transportation/Turnstile-Usage- Data-2019/xfn5-qji9
4https://data.cityofnewyork.us/City-Government/2020-Neighborhood-Tabulation-Areas-NTAs-
Tabular/9nt8-h7nd

D. Qiang and G. McKenzie 60:3
origins and destinations were combined to determine the final trip volume. For the subway
turnstile, the total number of entries and exits for every turnstile within a station was
summed. For example, there are four control areas in the “Cortlandt St.” station and each
control area has 10 turnstiles. The trip volume for that station was calculated as the sum of
all passengers through the 40 turnstiles. The trip data were then divided by the populations
of their respective neighborhoods. This process was straightforward for the bike-share and
subway turnstile data as they are represented as point geometries. The taxicab trip data,
however, is reported by polygonal taxi zone
5
(TZ). A dasymetric mapping approach was
used to allocate taxicab trip origins and destination TZ data to the NYC neighborhood
boundaries.
To address
RQ1
, our static
6
mobility vitality measure was generated by summing the
individual transportation mobility patterns across each region, producing a single value for
each neighborhood. While we took an “equal weights” approach here, the measure is designed
to allow a user to adjust the importance (weights) of each individual transportation mode
in the overall mobility vitality result, depending on their interests. Given this measure of
mobility vitality, we then examined how such a measure could be used to better understand the
vitality and variability of mobility services within a city such as NYC. To start, we averaged
the mobility vitality measure for each neighborhood by weekday and weekend. This allowed
us to subtract weekend mobility vitality from weekdays to better identify temporal variations
in mobility and differentiate neighborhoods based on prototypical commuting behavior.
Finally, we examined mobility vitality as a measure on which to identify similarities between
neighborhoods based purely on how inhabitants and visitors use different transportation
systems. To address this
RQ2
, we used Jensen-Shannon divergence (JSD), a method for
assessing the (dis)similarity of two probability distributions. In our case, we took the
trip volume for each day of the week of our three transportation modes as a distribution.
Having one distribution for each neighborhood allowed us to assess the similarity between
all neighborhood pairs. We then identified the neighborhoods that were most similar to all
other neighborhoods and those that were most unique.
3 Results and Discussion
For all three modes of transportation, there is greater mobility activity on weekdays than on
weekends. For bike-share origins and destination points, the population-normalized mean
values are 0.028 and 0.021, for weekdays and weekends, respectively. Similarly, the mean
population-normalized taxi pick-up density on weekdays is 0.0143, while on weekends it is
0.0137. The subway turnstile data was much more pronounced with a population-normalized
weekdays value of 1,407.86 and a weekend value of 839.91. These large values indicate that,
for many of the neighborhoods within NYC, the number of subway passengers is several
orders of magnitude higher than the residential population.
The weekday/weekend variation in normalized transportation trips is shown in Figure 1.
In order to compare weekday trips and weekend trips, we delineated the legend on the maps
by setting 0 as the dividing line in the class intervals. In both bike and taxi categories,
those values greater than 0 and those less than 0 were separately averaged into two intervals.
For the metro map, given the significantly higher number of weekday trips compared to
weekend ones, only one level was established for values less than 0, while those greater
5https://data.cityofnewyork.us/Transportation/NYC-Taxi- Zones/d3c5-ddgc
6Static here refers to the fact that temporal variability was not included in this approach.
GIScience 2023

60:4 Mobility Vitality
(a) Bike share. (b) Taxicab. (c) Subway turnstile.
Figure 1 Population-normalized weekend trip counts subtract from weekday trip counts, for
three modes of transportation.
than 0 were evenly divided into three levels. In general, the Manhattan business district
witnesses a predominance of weekday trips over weekend ones, with the intensity varying
across transportation modes. Bike sharing predominantly favors weekdays, with Central Park
being the exception. Conversely, neighborhoods encompassing recreational areas report higher
weekend bike trip volumes. Taxi trips exhibit a starkly distinct pattern, with higher weekend
volumes in both northern and southern Manhattan, notably in downtown neighborhoods
near Queens. Subway data, however, shows a universal weekday preference, except in East
Elmhurst and North Corona. A clear spatial clustering of neighborhoods with the greatest
discrepancy between weekday and weekend trips is evident in downtown Manhattan.
The results of the equally-weighted static mobility vitality measure are shown in Figure 2.
The operating region for the bike share service is the most spatially restrictive of our data
and so all data sets were restricted to this analysis area. As one can see, the greatest degree
of mobility vitality is in Central Park and the southeast corner of Manhattan. As one moves
towards the east side of Brooklyn and north Harlem, the vitality gradually decreases.
Figure 2 Static mobility vitality as calculated by summing the population-normalized trip volume
from three different modes of transportation.
The results of our Jensen-Shannon divergence approach are shown in Figure 3. In this
Figure, pink neighborhoods are the most unique neighborhoods in terms of mobility vitality,
reporting the highest average JSD values. These include the Upper West Side (Central),

D. Qiang and G. McKenzie 60:5
Upper East Side-Yorkville, East Midtown-Turtle Bay, Williamsburg, Harlem (North), and
Astoria (North)-Ditmars-Steinway. Among these, four are located in Manhattan, while
Queens and Brooklyn each contain one. Comparing these results to the static mobility
vitality map shown in Figure 2, it can be observed that the six most dissimilar neighborhoods
are not the ones with the highest static mobility vitality. They belong to the lower-scoring
group in terms of the three individual mobility patterns as well as static mobility vitality.
This speaks to the influence of the temporal dimension on assessing the vitality of a city
with respect to mobility. In our data, most neighborhoods with high static mobility vitality
have relatively low divergence values, indicating that they tend to be similar to one another.
Neighborhoods with low JSD values are in regions with high volumes of everyday traffic
for each mode of transportation and their individual mobility patterns show little variation.
The six most unique neighborhoods, as measured by our mobility patterns, are scattered
throughout the city but share a common characteristic, they are all waterfront neighborhoods.
Figure 3 Unique and similar neighborhoods as measured through three modes of transportation
using Jensen-Shannon divergence.
4 Conclusions and Next Steps
In this preliminary work, the concept of mobility vitality is proposed to measure the degree of
transportation and mobility within a region. This work investigates mobility vitality patterns
when different transportation starts or ends in the neighborhood and uses these patterns
to identify the divergence between different neighborhoods within NYC. Not surprisingly,
we found that mobility patterns are different on weekdays than on weekends. In most
cases, the volume of trips in downtown neighborhoods is greater during weekdays than on
weekends; however, taxicabs in some central business districts are the exception. Additionally,
neighborhoods with excessive divergence are dispersed and more dissimilar neighborhoods
often exhibit a high degree of clustering and high mobility vitality.
The next steps for this work will involve including additional modes of transit and
assessing the robustness of our approach through varying types of transportation. Our
current mobility vitality approach is meant as a “proof-of-concept” and further iterations will
GIScience 2023

60:6 Mobility Vitality
allow users to vary the weights depending on the question they are investigating. Last, the
current analysis was conducted using data collected at a daily temporal resolution. We aim to
examine the spatiotemporal characteristics of mobility vitality with a finer time granularity
in the future.
The results of the analysis presented in this short paper are meant to offer a glimpse
at the objective of generating a mobility vitality measure that represents the spatial and
temporal dynamics of mobility within a city. Through developing such a measure, our aim
is to empower urban and transportation planners with measures by which similarities and
differences within a city can be identified. Planners and government agencies will be able to
monitor how transportation policies can change vitality within a city and use such a measure
to improve equitable access to transportation systems within the urban environment.
References
1
Ezgi Eren and Volkan Emre Uz. A review on bike-sharing: The factors affecting bike-sharing
demand. Sustainable Cities and Society, 54:101882, 2020. doi:10.1016/j.scs.2019.101882.
2Jane Jacobs. The death and life of great American cities. Random House, New York, 1961.
3
Chaogui Kang, Dongwan Fan, and Hongzan Jiao. Validating activity, time, and space diversity
as essential components of urban vitality. Environment and Planning B: Urban Analytics and
City Science, 48:1180–1197, 2021. doi:10.1177/2399808320919771.
4
Qian Li, Caihui Cui, Feng Liu, Qirui Wu, Yadi Run, and Zhigang Han. Multidimensional Urban
Vitality on Streets: Spatial Patterns and Influence Factor Identification Using Multisource
Urban Data. ISPRS International Journal of Geo-Information, 11(1):2, 2022. doi:10.3390/
ijgi11010002.
5
Grant McKenzie and Daniel Romm. Measuring urban regional similarity through mobility
signatures. Computers, Environment and Urban Systems, 89:101684, 2021.
doi:10.1016/j.
compenvurbsys.2021.101684.
6
Patrizia Sulis, Ed Manley, Chen Zhong, and Michael Batty. Using mobility data as proxy
for measuring urban vitality. Journal of Spatial Information Science, 16:137–162, 2018.
doi:10.5311/JOSIS.2018.16.384.
7
Yang Xu, Dachi Chen, Xiaohu Zhang, Wei Tu, Yuanyang Chen, Yu Shen, and Carlo Ratti.
Unravel the landscape and pulses of cycling activities from a dockless bike-sharing system. Com-
puters, Environment and Urban Systems, 75:184–203, 2019.
doi:10.1016/j.compenvurbsys.
2019.02.002.
8
Yihong Yuan and Martin Raubal. Analyzing the distribution of human activity space from mo-
bile phone usage: an individual and urban-oriented study. International Journal of Geographical
Information Science, 30(8):1594–1621, 2016. doi:10.1080/13658816.2016.1143555.
9
Peng Zeng, Ming Wei, and Xiaoyang Liu. Investigating the Spatiotemporal Dynamics of
Urban Vitality Using Bicycle-Sharing Data. Sustainability, 106(1):1714, 2020.
doi:10.3390/
su12051714.
10
Bin Zhang, Shuyan Chen, Yongfeng Ma, Tiezhu Li, and Kun Tang. Analysis on spatiotemporal
urban mobility based on online car-hailing data. Journal of Transport Geography, 82:102568,
2020. doi:10.1016/j.jtrangeo.2019.102568.
11
Zhonghao Zhang, Yusi Zhang, Tian He, and Rui Xiao. Urban Vitality and its Influencing
Factors: Comparative Analysis Based on Taxi Trajectory Data. IEEE Journal of Selected
Topics in Applied Earth Observations and Remote Sensing, 15:5102–5114, 2022.
doi:10.1109/
JSTARS.2022.3183176.