Cartograms are maps that distort geographical shapes to convey their attributes. While less suitable for serious analysis than choropleths, they can be more visually engaging. In this mini-project, I used cartograms to visualize green and yellow cab ridership in New York City in April, 2019.

Method

1. Combine green and yellow cab trips into a single pandas dataframe.
2. Sum passenger pick-ups and drop-offs by zone, time (to the nearest half hour), and day type (weekend or weekday).
3. Calculate the cumulative sums for both pick-ups and drop-offs for each zone.
4. Adjust zone area according to the expression: 1 + (cumul. drop-offs - cumul. pick-ups) / cumul. pick-ups
5. Adjust zone color according to number of drop-offs at a given point in time.

Originally inspired by Yan Holtz’s work in R, this project uses Matthieu Viry’s cartogram_geopandas package and Paco Nathan’s approach to creating gif animations. To retrieve the taxi trip data, I also borrowed code from Chih-Ling Hsu’s blog post.

Results

As expected, the animations show that most drop-offs occur in mid-town Manhattan. Surprising, though, is that trip volumes are asymmetrical. There are more taxi trips going into the CBD than those leaving, which suggests that many commuters take a cab to work but then take transit back home.

I plan to recreate the animation when New York City releases data from 2020 in order to show how lock-down orders have affected taxi ridership.

Check out my Jupyter Notebook here.