Premise

We were tasked with planning an advertising strategy around NYC subway stations. The end result was a recommendation of NYC stations near which promoters should focus their efforts.

The ultimate goal was to raise awareness and participation for a fundraising event promoting women in technology.

Strategy

NYC’s subway usage data is readily available through the MTA website. This data was readily used to predict stations with high marketing potential.

In choosing stations to recommend, we evaluated the following:

  • Volume (how many people go through the station every day?)
  • Time of day
  • Weekdays or weekends
  • Proximity of technology companies

Results

Intuitively, public transit experiences higher volumes on weekdays than weekends, as people are commuting to/from work.

The peak windows for overall volume over the course of the day were during morning commutes (7-9AM) and evening commutes (4-6PM) (not shown).

Above shows the final recommendations for stations during morning weekday commutes. The black marks correspond to technology companies in the area, while the red pins are the stations themselves.

Recommendations
  • 47-50 STS ROCK
  • GRD CENTRL-42 ST
  • 34 ST-HERALD SQ
  • 23 ST
  • ASTOR PLACE
  • 18TH ST
  • LEXINGTON AV/53
  • TIMES SQ-42 ST
  • 42 ST-BRYANT PK
  • WALL ST

Technologies

Python

  • pandas
  • geopy
  • matplotlib

Google Maps API

StackOverflow.com

Thoughts

  • Data processing is much faster in pandas than it is in loops and dictionaries.
  • Perfectionism will get you nowhere.
  • Bare minimalism will get you nowhere either.