Introduction: Empty Taxi Trips

Chicago, Illinois is the third largest city in America, boasting a resident population totaling nearly three million people spread over 234 square miles. The city consistently ranks among the top 10 most visited cities in the country and has not one, but two international airports serving the city, allowing for around 55 million tourists to visit Chicago annually. With such a large influx of people every year due to business and tourism, along with the people of Chicago who already call the city home, there is a constant demand for highly efficient modes of transportation into and around the city. Arguably, the city of Chicago’s most iconic and versatile means of transportation is, and remains, the Taxi.

For a taxi company where efficiency is a necessity in maintaining profitability, understanding how to maximize their fleet’s cost of time is essential for successful business operations. Studying taxi trip data, made publicly available by the City of Chicago, we venture to explore taxi utilization, specifically empty trips, where taxis have no passengers. By analyzing empty trip data, or time spent off the clock, we can gain better insight as to the rhythms and patterns of Chicago's taxi services and make more educated decisions regarding transportation and city-wide mobility.

Observation and Findings

Our approach to work through the complexity of this condition was to first identify the empty taxi trips by analyzing the time spent between pickups and drop offs for each individual taxi in our dataset. This time was considered “idle tIme,” where taxis were not actively transporting passengers; this time ranges from time spent taking a break, to driving to another location, to taking time off of work.

After filtering our dataset and restructuring it through visual representation, we were able to observe clusters and peaks in our graphs, which led to some interesting findings. Using these visualizations, we are able to offer an educated speculation on how the average taxi driver in Chicago utilizes their time. Being able to better understand the dynamics of cost of time allow for increased profit amongst taxi companies Our visualizations display our data in two different graphic forms, which represents frequency of empty taxi trip time plotted throughout a 24 hour time period and also weekly.

Analyzing idle trips during a 24-hour period

In the streamgraphs it can be observed that there is a large variation in time of empty trips starting around 7 AM, then peaks around 6pm with the greatest amount of empty trim time amounting to less than 30 minutes on average. Since typical weekday weekdays start around this time in the morning, we can speculate that this peak shows us typical taxi traffic transportation patterns for morning commutes, whereby taxis are frequently picking up and dropping off customers.

Another peak of increased frequency of empty trips occurs around the evening time at about 6 PM, which might indicate an end to the typical workday or a lull between the workday and evening leisure activities.

Analyzing idle trips over a weekly period

The colors indicate the days of the week (Sunday - Saturday)

Choose time: Toggle sort

When scrolling through late evening and early morning hours, we can see a drastic difference between Friday and Saturday night versus the rest of the week. We believe this high amount of short empty trips is due to people going out to bars and clubs.

Choose time: Toggle sort

The graph shows three peaks, at 15 hours, 40 hours, and 64 hours. The peak at 15 hours is very intuitive as this would be the time spent off-duty. There appears to be no interesting patterns in terms of the day of the week these idle times happen. However this is not the case for the 40 hour and 64 hour peek.
Choose time: Toggle sort

When looking at the totals, we can see a high amount of about 40 hours of idle trips ending on Sundays and Mondays. We believe that when taxi drivers choose to take one day off a week, they either choose Saturday or Sunday. Also when analyzing different times, we see that when taxi drivers take Saturday off, they tend to start work around 10AM on Sunday. When drivers take Sunday off, they tend to start at around 7AM on Monday. For idle trips around 64 hours, we see most of these trips end on Monday which suggests that when taxi drivers take two days off in a row, they tend to choose Saturday and Sunday.

Conclusions and Moving Forward

The research discussed in this paper sought to further analyze and understand the complexities behind taxi operations in a large metropolitan city. Using publically available data from the City of Chicago, our team analyzed lengths of idle taxi trips, where operators had no passengers. Through the use of computational analysis paired with data visualizations, our team was able to gain interesting insights into the weekly operations of Chicago’s taxi system.

From our visualizations, we were able to observe accumulations of idle trip lengths plotted across a week. We found that there are two peaks of these idle trip accumulations that occur at the beginning and end of the work day, 7 AM, and 6 PM respectively. It is plausible the reason for these increased accumulations of idle trip lengths might have to do with traffic congestion occurring during these peak traffic hours and drivers not being able to travel efficiently. Looking again at the streamgraph, an additional accumulation of idle trip lengths occurs around hour 40 and 64 out of a 168 hour work week (starting on Sunday). These blocks of idle trips occurring over the weekends represent time spent between the last passenger drop off to the next passenger pickup. From our dataset we can speculate that when taxi drivers choose to take one day off a week. They either choose Saturday or Sunday.

Our research has led us to a better understanding of Chicago’s taxi operations. Expanding on this research, it would be interesting to see how these idle trip times could be expressed in terms of cost of time and missed opportunities to analyze what economic impacts idle trips have on a taxi company’s revenue. Assuming the lengths of idle trips could be converted into monetary terms (time, fuel spent, overhead, driver’s wages, wasted opportunities, etc.) economic costs would become clear, and solutions for reducing wasted idle time would likely become a priority for profit-driven taxi companies.

Team & Contributions

Davis Millarrd

Helped outline the organization of the webpage, assisted Connor in primary interpretation of the visual graphical analysis’ resulting in our findings and observations, wrote and edited the narrative text displayed in each category (Introduction: Empty Taxi Trips, Observations + Findings, Conclusion + Moving Forward), and assisted in finding sourced images to use for overall web page aesthetic. Ensured the team kept pace with work deadlines.

Connor Burdick

Cleaned the data from our data source using Python. Created the streamgraphs and grouped bar charts using d3. Helped format webpage. Helped discuss narrative and findings from the graphs.

Aaron Perrill

Wrote and edited the website text specifically: “Introduction: Empty Taxi Trips”, “Observations + Findings”, “Conclusion + Moving Forward”. Organized all virtual team meetings and guided group discussions. Helped in acquiring images for website aesthetics. Helped in guiding the general and aesthetic organization of the website.

Mohammad Fasahat

Created a new GitHub repository to launch the webpage. Built the animation effect for the introduction of the webpage and designed the frame. Edited the HTML and CSS files. Helped by discussing and researching the narratives and graphs.