Methodology

With the normality assumption for the population distribution of Citi Bike trip durations, I am testing a single variable (i.e., trip duration) between two categories within an unpaired data set that has 285,552 observations. According to these criteria, I chose to perform Welch's t-test, which compares the difference in trip duration and test if the difference is significant at the designated significance level of 0.05.
My peer Srikanth also suggests conducting either a z-test or Welch t-test depending on the homogeneity of variance and under the assumption that the data is parametric. I adopt the t-test with the consideration that there may be differences between the variances of two groups.
I thus conduct a one-tail test under the null hypothesis that the average trip duration on weekends is the same as or shorter than that of weekdays, on which hasty commuting takes a greater proportion.

Conclusions

Both Fig 2 and Fig 3 show that there is seemingly little difference among each day of week and between weekdays and weekends regarding the average trip duration, which is around 11 minutes. The result of the t-test also shows no statistical significance for the difference. The t statistic for the test is 0.65, which is far smaller than the threshold of 1.646 (one-tail, alpha = 0.05). I can not reject the null hypothesis nor conclude that weekend trips are longer than those of weekdays in average.