This week, I will be comparing trip distance as well as trip duration based on different groups and different times when the sample was collected.
Day of the week
In the past week, I found moderate correlation between the number of trips and time of the day, so this time I decided to explore how time variable affects the outcome in other ways. I will be performing an ANOVA test to find out how the mean differs for samples taken on different days.
I will perform two separate ANOVA tests, one for trip duration and another for trip distance. I am performing this on commute trips rather than leisure because commute trips contain less outliers (they don’t include round trips, some of which are just someone who took out a bike, changed their mind and docked it). This analysis will compare rider behavior across different days of the week.
Effect on trip duration
To remove outliers, I only considered trips under 500 minutes (or 8.33 hours) before performing the test.
Effect on trip distance
For distance analysis, I excluded distances that are zero.
According to the ANOVA analysis, both variables did not vary significantly across different days.
Rider Type
Next, I will perform a t-test to compare trip durations for casual riders and members and see how their riding behaviors differ. Again, I will be filtering for commute trips.
Effect on trip duration
Since the p value is virtually zero, that means there is a statistically significant difference in trip durations between casual users and members, with casual users having longer trip durations on average.
The mean trip duration for the casual group is 13.68098 and for the member group is 11.13364. Since the mean of the casual group is higher, it suggests that casual users tend to opt for using Bluebikes for longer trip durations than members.
Effect on trip distance
Again, results show a statistically significant difference between the two groups (p=~ 0), with casual riders going on longer and farther away trips. Casual riders had trips with an average distance of 2.499 kilometers, where members averaged 2.094 kilometers.
Conclusion
Rider type (member or single-time user) had a more significant effect on the outcome for trip durations and distance, compared to time of the week which had little significance. Members averaged shorter trips than casual riders, both in terms of duration and distance. This could indicate that casual riders are only willing to pay the fee for a single ride when the trip is long enough to be worth it, while members use bikes more regularly even for much shorter trips.
Seeing Boston Neighborhoods through Administrative Data 
Hi zmarhoon1,
Great post. Your assignment found that ‘Rider type (member or single-time user) had a more significant effect on the outcome for trip durations and distance, compared to time of the week which had little significance. Members averaged shorter trips than casual riders, both in terms of duration and distance.’ It’s possible that casual riders only pay for trips when they need to. Members pay monthly or yearly and tend to use the bikes more frequently. Blue bikes guarantee unlimited 45-minute trips for members, so it makes sense to use them more frequently.
LikeLike