WHO RIDES THE BUS: Examining Transit Ridership in Marion County WHO RIDES THE BUS | Page 5
meaningful groups using the natural structure
of a dataset. In this type of analysis, there
is no single “true” grouping [2]. Instead we
identified groups of riders by clustering them
based on selected attributes. As a result, a
respondent in one group is more similar to
their groupmates than to members of other
groups. These naturally occurring segments
provide a useful abstraction from individual
records in the dataset, which allows us to
analyze and summarize the data based on the
qualities that group members share.
… Provides Affordable Transportation Options to
Riders Living in Areas with High Housing Costs
Across all groups, IndyGo riders tend to live
where housing costs are high in proportion
to their incomes. Affordable transportation
options are necessary to keep these residents
in their home neighborhoods which in turn
contributes to community stability. Some
riders, such as college students in Groups B
and C, may forego moving to a less expensive
neighborhood farther from their classes if
it means sacrificing access to a variety of
transportation options, including transit.
Similarly, commuters – especially in Group
A – who rely on the existing transit service in
their neighborhoods may be unwilling to move
to less expensive housing markets with lower
transit service, because their transportation
costs could go up.
Our app roach to clustering is user-centric;
that is, we were more interested in grouping
riders based on their personal characteristics
than the nature of their transit trip. So our
analysis included three attributes about
riders themselves and one attribute relating
to the availability of transit in their home
neighborhoods. The metric we used for
transit service is revenue miles per square
mile: the number of miles an in-service bus
drives through a neighborhood, per square
mile of area in that neighborhood. IndyVitals
[3] and neighborhood areas were used as the
boundaries for this analysis.
How Rider Groups were Developed
The profiles on the following pages summarize
the results of a cluster analysis that yielded
five meaningful groups of riders. We also
mapped riders to their home neighborhoods
to further explore the relationship between
rider behavior, transit use, and the built
environment, therefore providing a general
audience with an informed geographic
approach to the transit survey dataset.
The preliminary analysis yielded eight groups
of riders. While verifying the rider clusters, we
combined similar groups into five meaningful
superclusters to better reflect the riders in
our community. A discussion of the technical
approach is provided in Appendix A.
We generated the rider profiles using cluster
analysis, a flexible approach to creating
Marion County Transit Plan
The IndyGo Forward plan [4], developed through public involvement, shifted transit
resources from coverage to ridership. Focusing resources on ridership means running
frequent transit in high-density areas with high ridership. Coverage instead spreads
service across an entire community. With greater coverage, routes are less frequent,
but access to transit is greater.
The Marion County Transit Plan achieves a more ridership-focused system by adding
more high frequency lines (where vehicles arrive every 15 minutes) in high-ridership
areas and creating three rapid transit lines (where vehicles arrive every ten minutes).
This report serves as a benchmark for ridership characteristics before these changes are
implemented, and as one possible framework to view how these service improvements
will impact current ridership.
5