Content uploaded by Hung Cao
Author content
All content in this area was uploaded by Hung Cao on Mar 24, 2019
Content may be subject to copyright.
How can graph databases improve transit systems?
Hung Cao, Ikechukwu Maduako, Emerson Cavalheri, Ryan Brideau, Monica Wachowicz
{hcao3, imaduako, e.cavalheri, Ryan.Brideau, monicaw}@unb.ca
People in Motion Lab, University of New Brunswick
CODIAC Transpo currently operates 30
regular routes Monday to Saturday,
some of which provide additional
evening and Sunday services. The map
shows the 400m catchment areas of all
bus stops in Moncton.
Motivation
Conclusions
This research project has developed a graph database model to provide easy-to-access information on transit options, incentivize the use of transit, and explore
different transport capacities. Graph databases are unique and novel because they provide different stakeholders with a new set of policy education, exploration, and
management options. The Black Arcs anticipates civic engagement tools as a key component to further explore the use of graph databases in transit planning.
Graph Database Design
This chart compares the actual arrival time to the expected arrival time by
hour of day for a single bus route. It uses one week of data, from June 3rd
to June 9th, and compares the arrival times for every stop along the route.
At times of day where there is lower ridership, the charts skew toward
early arrivals since the buses do not pause at the stops without riders,
causing them to be early for the following stops.
•Collect
data
Step 1:
•Filter data into
the group of
data files
Step 2: •Compute the Move
and the Stop status of
each data row
Step 3:
•Do the
annotation for
each data row
Step 4: •Compute street segments
and do annotation on each
street segment
Step 5:
•Index Trips and compute
Arrival Time and
Departure Time
Step 6:
The graph database contains the
geographical location of each bus every 5
seconds during a period of 2 weeks.
Approximately 900,000 nodes and 4.5
million relationships have been created
for the 30 regular routes.
The trip connectivity graph above
shows one specific trip taken by abus.
The nodes of the network represent
the actual stops and moves that have
occurred during this trip, as well as
their geographical location type (e.g.
the street segment, bus stop, or street
intersection).
The degree centrality graphs below are
showing the nodes with the highest
number of connections. The graph
database allow us to retrieve the
busiest streets and bus stops in the
network according to the centrality
measure. The bus stops located at the
Plaza is busiest bus stop and Main
street is the busiest street in the
network. The graphs above show the longest
and shortest trips over two weeks. We
have used the longest and shortest
path queries to count the number
edges for all trips and find the longest
and shortest trips by hour. It can be
seen that the number of edges vary
with time, space and traffic volume.
The metropolitan area
of Greater Moncton is
the fastest growing census metropolitan area in eastern Canada. We are
working with CODIAC transit to promote efficient transportation services and
reduce private car dependency.
Queries Data Visualization