This is an interactive R/Shiny dashboard that displays an interactive ridership summary of the Massachusetts Bay Transportation Authority transit system in Boston, Massachusetts.
The application is deployed using shinyapps.io, and can be viewed here.
The application was built using the R web application development package called Shiny. Shiny is an open source R package developed by Posit with the aim of helping data scientists and analysts communicate statistical findings by creating complex statistical web applications.
The dashboard uses publicly available data, found on the MBTA open data portal, Blue Book. Average weekday ridership data for bus routes, rapid transit routes, and ferry routes, as well as all of the stops along these routes are available for the Fall 2018 and Fall 2019 season across the three modes of transportation.
Data for some modes of transportation is available for an additional number of years, either before or after the 2018-2019 range, but the data availability was not consistent, so only data from 2018-2019 was included.
This repository contains a cleaned, aggregated version of the data, but the original raw form is available at the Ridership directory of the above link.
In addition to ridership data, shapefiles for the geography of each route and stop are also available at the data portal, specifically in the GTFS directory.
These files were used to generate customized shape files, with the aid of the sf R package. Many routes had the same ids in the original shape files, since portions of each route
were separated as per the MBTA file convention. These custom files differ from the original shape files only in that portions of each route were attached such that every route
had a unique id, and each route comprised of each of its respective sections to form a continuous line on a map.
An exception to this is the case of the MBTA's Green Line, which is actually comprised of the B, C, D, and E routes of the Green Line. These are in fact separate routes, however, data from the Ridership link above generalized ridership to the sum of its parts (B, C, D, and E). Therefore, the choice was made to connect each of these subroutes to form one, continuous line.
The result of this action can be seen in the image below, where all 4 sections form a single Multilinestring (to use Well-Known-Text terminology).
To ensure that the dashboard is stable and easily re-deployed given any number of feature or styling changes, its package dependencies were tracked
using the renv R package. This creates a json-like file, titled renv.lock, which stores information about each dependency, its version, and subsequent dependencies.
Shinyapps.io ensured base linux dependencies were managed before deploying to a server, however a dockerfile could easily be used to track this information as well in the case of deploying to, for example, an AWS EC2 instance. This is my typical approach, since I typically deploy my apps to EC2 instances. Shinyapps.io was chosen this time because I wanted to explore other options.
The total amount of time invested in creating this application was about 7 days, much of which was spent finding the data, examining variable definitions, and cleaning the data.
Thank you to Appsilon for creting the rhino R/Shiny framework, which was used to create the application. Additionally, thank your for creating the Appsilon template
bundle, which greatly inspired the design choices used to create the app.