Bike trip duration is affected significantly by traffic, weather conditions, time of the day, user type and so on. However I realized that Chicago's bike-sharing platform Divvy gives a fixed estimated travel time between two docks.
The scatter plot displays the average travel time on the vertical axes and number of trips taken on the horizontal axes for each hour of a day. Please hover over the circles in order to see what time slot they represent. The busiest hours are 17,8,16,18 which are known as rush hours. As number of trips increases average trip duration becomes longer supporting our hypothesis that trip durations differ with respect to a set of parameters one of which is the time trip is made.
The following map shows the most 100 common bike routes based on Divvy data. Colors and sizes represent number of trips between each trip of docks.
The histogram shows trip durations for one of the most common routes, which is from 'Theater on the Lake' to 'Lakefront Trail & Bryn Mawr Ave'. Average trip duration between between 9am-10am is 26.1 min, it is 29.7 min overall and 33.5 min between 7pm-8pm. T-statistic comparing the means for 9am-10am and 7pm-8pm is 2.79 and resulting p-value is 0.006 so the means are different. The difference is critical because single-ride limit is 30 min so users have to pay extra amount for every minute they exceed 30 min.
My estimation of trip durations using Random Forest regression decreases the MSE by 92%.