Introduction

For this report, we need to explore a certain topic in the form of a map, so New York City – the city where I now live – was my first choice. After browsing through many databases on the NYC planning website, I focused on the topic of bicycles and their users in New York City for my research. In this lab report, I will study the relationship between existing bike lanes in Manhattan and bike traffic at selected points and will compare the traffic at each point on weekdays versus weekends. Furthermore, I will study Manhattan bicycle usage’s growing trends as well.

Dataset

First, I was inspired by a database of Manhattan bike counts from the New York City Department of City Planning’s (NYCDCP) Transportation Division.

It records the number of bicycles at 15 locations in Manhattan between 2010-2015, as well as the number of users, year of use, and use of bike lanes, and also records the flow of traffic at the locations separately on weekdays and weekends, which also provides me with a convenient way to compare bike use on weekdays and weekends.

In addition, I wanted to reference it to other databases, and I keyed in on “New York City” and “bicycle” and found an open database of existing bike lanes in New York City. This is an annual update of bike lane routes in the city by NYC Open Data, which serves as one of the layers of the map and provides me with a reference for comparing information.

Inspiration

The visualization was inspired by the map displayed on Citibike when looking up nearby bikes (Reference1), where the routes would be shown as lines and the bike locations as points on the map. I decided to borrow this format and use it in my lab report.

The second inspiration for the reference was a past student’s work in which the trend of the Apple map was displayed on a sheet by the time change. I thought this had similarities to my research topic, and since I also had the year change in my database, I could apply it to my lab report. (Reference2)

Tool

The visualization tool used in this report is Tableau public, the same as in report2. Tableau public is an easy-to-use software with a wide range of visualization options, and it can meet the visualization needs of most geographic files, automatically generating visual maps based on the location data and information provided.

The gif tool used in this project is Photoshop, the timeline function can help me easily create the animation effect based on certain still pictures.

Process

In the first step, I needed to process an Excel database containing bicycle data for 15 points, consisting of three sheets that recorded bicycle traffic on different routes and at different times. (Fig1-1) I merged them into one big sheet in Excel according to “route type” and “time type” and then saved them in XML format. (Fig1-2)

The downloaded bicycle route database is a spatial file that can be dragged directly in to Tableau Public for use. After dragging the geometry data into the created sheet, the following chart is obtained (fig2-1). You can see that the map of New York City is covered by the identified bike routes.

After checking the data details, I found that these routes are actually made up of three different types, so I chose to use bike lanes to filter color and set different shades of the same color for them to serve as a reference layer without disturbing the user’s attention. (figure2-2) We can infer from the visualization that the more colorful the area, the higher the overlap of bike lanes, and the more dense the distribution of lines, the more convenient the traffic development. The light blue lines are non-trunk roads, which are represented as “null” in this report, so they imply fewer bicycle route possibilities. The above inference also provides information for the subsequent comparison of the data.

To project the point locations from the XML database onto this map, I need to create points that the Tableau public can recognize and associate with the spatial database described above. I observed that the XML file contains the longitude and latitude for identifying geographic information. Therefore, I dragged the XML file into the add source screen and used the command “make point,” selecting “intersect” and “Full outer” in the relationship. “, because I want to find the overlapping area and relationship between the two while keeping the existing layers. The specific operation is shown in the figure below (fig3).

Thus, the basic database for my visualization has all been imported successfully. (fig4)Next, I want to visualize the specific data of the report. First, I studied the relationship between bicycle flow values at several points in Manhattan, and the bicycle route map was only used as a reference, so I locked the bike lane layer in the layers panel and turned off the interactivity feature of this layer. (fig5) At the same time, I reduced the transparency of the bike lane layer to 65% to highlight the traffic situation at the points more.

While working with the bicycle traffic at the point, I observed that the main items available for comparison in the table were the total number of users, the period of use, and the year. Therefore, I decided to set these items as Mark and Filter tools to complete the visualization.

First, I tried to mark the total number of users according to the shape and added red color to the points (to distinguish them from the background color) to get the following figure (fig6). We can see in the figure: that the two spots along the river area in western Manhattan have the darkest color and the largest area, with the number of users at 95235 and 62116, which indicates that the bike traffic in western New York is higher than that in the east. At the same time, we can see that most of the dots are concentrated in Lower Manhattan. I speculate that this is because the subway lines in Lower Manhattan are not as abundant as those in Midtown and Uptown, and more Asians are living in Lower Manhattan, making bicycles their most preferred mode of travel. At each location, the total number of users reached over 10,000.

Also, try to compare these circles with the route map layers, and we will not be surprised to find that these points are often located where multiple routes overlap or intersect. This is especially true in the case of the lower urban area, where the concentration of points is also the densest in terms of bicycle routes. Therefore, we can tentatively infer that the number of bicycle users is positively related to the density of bicycle routes in the area.

The second attempt is the filter the usage period. To compare the spot traffic on weekdays and weekends, I set a filter in the right sidebar, including two options: “weekday” and “weekend.” The user can check the box to understand the relationship between usage time period and spot traffic. The images of weekends (fig7-1) and weekdays (fig7-2) are shown below.

Among them, we can see: that there are more active spots on weekdays than on weekends, which could be due to the traffic jams in the morning and evening rush hours on weekdays that make many users choose to bike.

The last attempt is about the year of use. Since the database records change between 2005 and 2015, I want to study NYC users’ bicycle use trends over the years. I converted the default “year” data type to string so that the tableau public would recognize it as a filter (fig8-1), and users would be able to change the year in the drop-down menu to see how bike traffic changed at a given point. (fig8-2)

I believe the drop-down menu feature in Filter on Tableau provides a nice timeline, but I think a gif is a better visual representation of how bike traffic changes at various points over the years. Therefore, I created a GIF using Photoshop which shows the trends in bicycle usage in order from 2005-2015. (gif1)

By mapping the GIF data, we can see that: between 2005-2006, only the west side of Midtown had higher bike traffic at point 14, while all other points had usage below 1000. From 2007 onward, Lower Manhattan grew rapidly, as reflected by the fact that the emerging Point 12 surpassed the original Point 14 usage in the first year of development; meanwhile, usage at other points in Lower Manhattan also steadily increased. After that, bicycle use increased year over year at most Manhattan locations. Starting in 2012, use at the Uptown and Midtown locations began to decline, and the city’s bicycle use nucleus gradually moved closer to the Downtown area.
This may be because MTA routes in the lower urban areas are not as popular as other parts of the city, while bikes can take people to more places they want to go.

Finally, I adjusted the details of the visualization report, such as the title, the shade mode of the background layer, the data color, and the label’s font, and exported it to Tableau public to get the following link.

Number of bicycle uses at 15 locations in Manhattan from 2005-2015

Reflection and future thoughts

This is my first attempt at combining GIFs with Tableau Public for data visualization, so the detail part leaves much to be desired. In this report, I simply used the time span as a filter to compare weekday versus weekend bike usage and analyzed the trends in bike traffic in Manhattan between 2005-2015. I observed other variables in the XML file that could be analyzed, such as user gender, and if more time were available, I would have analyzed this data as well.

Also, I noticed that the shades of colors sometimes do not simply and clearly illustrate specific values, so I still kept the labels section in the interactive links, but these labels inevitably overlap when they are adjacent, which can reduce the readability of the visualization report. tableau public is not very friendly to these detailed sections. If I continue improving this visualization report, I will explore other tools to improve these sections.