Fig.3 show the distribution of Citi Bike trips on riding durations.
Obviously, the number of short trips is about triple the number of long trips when we separate the trips by the day of week. Chi-square test also gives the same result that the null hypothesis is rejected on significance level 0.05. The chi-square statistics is 183.5 , which is far higher than the threshold of 3.84 (one-tail, alpha = 0.05, degree of freedom = 1). The result matches the assumption of the “last mile” theory that most of people regard Citi Bike as a commuter for short distance/time. However, there are some weaknesses in the analysis.
The data I used is the trips in January2014. It’s very likely that people chose not to ride for too long time in cold winter. The better way is to choose the data in each season in order to ensure that there is no bias in data selection.