The Tableau Project: Bike Sharing and Weather Visualization
This article was written by Alparslan Mesri and Hakan Elbas.
In this article, we will show you how to analyze and visualize data with Tableau.
We will explore data on shared bicycles in the Washington DC area based on summer 2011. Especially, we’ll look at how weather, day of the week, and member type affect usage.
If you’re ready, let’s start…
Tableau is a super-powerful data analysis and data visualization tool. It has been widely adopted by companies of all types to help them become more data-driven. It is an essential program that many data scientists, business analysts, and managers can use.
There are more than 4,300 community bikes in the Metro DC area.
Data
You can access bicycle shared usage data free of charge here. You can access the data about the amount of precipitation, which we will interpret together with the bicycle common usage data, free of charge here.
As a result of data cleaning in the Python environment, our data file has the following columns:
Duration: It represents the travel time.
Start Date: Travel start date.
End Date: Travel end date.
Member Type: It means that the bicycle user is a registered (Annual member, Special member, Special member for the day) or visitor type (Single-use, 24-hour, 3-day, 5-day) member.
HPCP: Shows the hourly precipitation amount.
When we look at only June, July, and August 2011, there are around 1.2 million records. We will therefore limit the study to these three months for this exercise. To make it easier to work with duration, you can create a calculated column after entering your Tableau worksheet. If you want to work in minutes instead of seconds, the time should be rounded to 60. We did this using Python. You can also do this in advance in Excel or Google Sheets if you wish.
I got precipitation data for Reagan National Airport (DCA) in NOAA from June 1, 2011 to August 31, 2011. After taking the necessary actions, NOAA will send you an email when the data is ready for download. The process only takes a minute. We have deleted all the columns in the csv file except precipitation and date using Python.
Data Visualization and Analysis
We have made several worksheets with visualizations that examine various aspects of the data. Then we combined several of them into one — — control panel, and it turned out to be a story, as you can see here. Since it is not possible to transfer the tables to medium, we took screenshots as follows.
In this article, we have provided only a small introduction to this deep field of visualization. The advanced features of the Tableau program that we appreciate are available in the full version. One of them is the opportunity to publish your graphics interactively on the web. If you wish, you can search this feature and open your own graphics to access with the server.
Our latest chart has a biaxial format. In general, biaxial graphs are not very helpful at first in giving the researcher an insight into the whole picture, but they can be very useful graphs in terms of showing the relationship between units of measure. In this final chart, the blue lines show the total cycling time, while the orange lines show the average rainfall.
Analysis
In Graph-4, high precipitation results in low utilization rate in some cases. There seems to be an inverse relationship between the amount of rain and the amount of cycling time.
It would be interesting to split the day into hours when looking at cycling amounts. Rain late at night or early in the morning probably has little effect on riding. We will consider this analysis in our later work.
In Chart-1, it’s surprising to see how much less bike use is on Mondays and Tuesdays in summer compared to other days of the week. This chart shows the weekday-use relationship very clearly.
The data in Graph-3 showing that users who have a member card cycle for much shorter periods of time than users who do not have a membership card may be found interesting by researchers. This can be explained by the fact that members mostly use bicycles to commute, while non-member drivers mostly use bicycles for touristic purposes. Of course, the accuracy of this argument can be examined by doing more research on this subject with more data.
In Graph-2, it can be observed that the total cycling time of members and non-members is almost the same. However, the average cycling time of non-members in Chart-3 was much higher. This gives us the information that members have a much higher amount of cycling.
Conclusion
In this article, using Tableau and combining the two datasets, we found some interesting information about shared bikes in Washington DC. Tableau makes data analysis fast; It took only a few hours to find the data, clean up, defragment, build the control panel and write this article.
In addition to the ones mentioned above, many more analyzes and projects can be done using Tableau’s other features. We performed this analysis using only 2011 data. Looking at the data of all years to see the big picture would have given us more interesting findings.
It would also be interesting to generate a trip forecast using machine learning. Tableau Public doesn’t allow for easy links to python code, but Tableau Desktop, which recently jumped to $70 per user, does.
