April 21, 2017

Are bikeshare trips being linked with L train travel in Chicago?

 

Recently, I have been studying the bikeshare systems in the Washington D.C. and Chicago areas. One of my projects focuses on the interaction between bikesharing and the L Train stations in the Chicago area, in particular, how L ridership is affected by the presence of bikesharing. When doing the analysis our results suggested that bikeshare is a complementary good to the trains only when looking at bikeshare trips ending near a L train station. When looking at trips leaving the area near train stations, called origin trips, our results showed a substitutional relationship between the two, with more origin trips being associated with fewer train trips. However, ridership data is only captured when someone is going into a station, and not when someone is leaving a station. For this reason, when analyzing the origin bike trips in the region, you are comparing them to trips of people who are leaving from that station, on the train. Origin trips means people are going away from the station and so it should be compared to when people are leaving a train station, not when people are going to it. With this in mind, I became curious about whether or not scheduled train arrival times are correlated to origin bike trips.

To make this analysis simple, I will only focus on one of the train stations used in my previous analysis, discussed above. The train station chosen was the Polk station on the Pink line, because it has bikeshare right near the train station and there is only one train that goes through the station. Unfortunately, I was unable to find historic arrival times for the station, so the scheduled arrival times will have to do. Because the train goes in both directions, both the train times going to and from downtown are included. With these times, I then found the difference for each directional schedule to all the ride times for a single work week (5/2/16-5/6/16) of data at the nearest Divvy bike station. After this was done, histograms of the differences were created in order to view the frequency of each time difference.

This first histogram looks at the amount of time between when a train arrived from downtown, and the start of a bikeshare trip.

From this first figure we can see that there are about 13 “extra” trips where the difference is 2 minutes or less.  This helps support the idea that people are linking the two modes, because it is greater than the rest of the values, which we would expect to be uniformly distributed.

This second histogram looks at the difference between train times going toward downtown, and the bike trip start times.

This second figure also supports the idea that people are linking the trips, because the frequency of bike trips that occurs within 2 minutes or less of the previous train is almost double the next closest frequency.

As shown in the graphics, the difference of 1 and 2 minutes was the most common difference between train arrival times and bike departure times. Although the train schedule data is an imperfect measure of arrival times, and the observation period is only one work week, it is likely that people are linking these two modes together at this station.  It is important to gain a better understanding of how these modes interact, because of the benefits that they can bring. On top to this understanding where to place bikeshare stations is important so that bikeshare can be successful. After doing this preliminary analysis, I have decided to dive further into this topic.