Based on the project’s business use-case, EDA was performed using the processed bike share ridership data in order to provide the following

  1. insights about user behavior based on temporal and geospatial bike share ridership patterns
  2. recommendations for which and when stations should be used during the campaign

🔂 High-Level Network Performance

The following exploration of the processed data was performed in order to understand the high-level historical performance of the network and its footprint

  1. yearly totals for

    1. trips (bike share ridership)
    2. number of stations
    3. number of bikes

    used in historical bike share ridership (2018 to 2022) and during the planned expansion window (2023 to 2025).

👤 Get Insights About User Behavior from Ridership Patterns

In order to address both requirements of the ‣ for this project, the following EDA was performed

  1. Identify top-performing stations
    1. the aggregated station demand (total bike share ridership) metrics used to classify stations as top-performers is discussed in the next section
  2. Extract insights about the attributes of both types of stations (top-performing stations and other stations)
    1. fraction of stations located in downtown Toronto, which dominated the original footprint of the bike sharing service in Toronto when it was launched
    2. fraction of stations located in
      1. downtown and adjacent neighborhoods (immediately East and West of downtown)
      2. other neighborhoods
    3. fraction of stations that accept a credit card as a method of payment
    4. breakdown of stations based on their physical configuration
      1. regular
      2. charging (supports e-bikes)
      3. stations with and without a dedicated check-in and check-out kiosk
  3. Extract insights from temporal patterns in user ridership trends for both types of bike share stations and for both types of bike share members (Annual and Casual), over the period from January 1, 2018 to March 31, 2023
    1. by month of year
    2. by hour of day per month of year
    3. by day of the week per month of the year
    4. relationship between daily maximum temperature and daily bike share ridership
  4. Extract insights from geospatial patterns in both types of stations
    1. by proximity of top-performing stations to neighborhoods within Downtown Toronto

These insights were used to understand attributes of the top-performing stations and to use the temporal and geospatial patterns to recommend which stations to select for use in the campaign and when displaying of digital ads should be prioritized at the selected stations.

🔝 Identifying Top-Performing Stations

đź’  Metrics for Identifying Top-Performing Stations

In order to recommend top-performing stations which should be prioritized for displaying digital ads, the following station performance metrics were used

  1. total departures during the most recent full year of historical ridership (2022)
  2. total arrivals during the most recent full year of historical ridership (2022)
  3. total departures during all full years of historical ridership combined (2018 - 2022)
  4. total arrivals during all full years of historical ridership combined (2018 - 2022)