Step by Step Building a Vacancy Tracker Using Tableau
Step-by-step explanations of vacancies valued in tens of millions of dollars in the small town of Fitchburg, Massachusetts.
By Dotun Opasina, Associate Data Scientist
Downtown Fitchburg. Source: City Government
Over the summer of 2021, I volunteered my data science skills to help Fitchburg’s City Hall planning office to build a data visualization to track business vacancies in the downtown area. The income generated from rent or the selling of properties are tens of millions dollars, and have significant importance to many stakeholders in our community.
For those that do not know, data visualization is an important aspect of analytics and there are many reasons for using visualizations to communicate business needs.
For this project, I used a cloud database, Tableau for visualization, and Python to perform data wrangling. Here are the steps below:
There were 5 steps used for this particular problem based of the data visualization process.
- Problem Framing with Stakeholders
- Storefront Vacancy Data Collection
- Vacancy collection
- Image capturing
- Data Storage
- Data Cleaning and Connection
- Geo-coordinates of location
- Mapping addresses to images
- Sourcing missing ownership information
- Dashboard developments using Tableau
Step 1: Problem Framing
The executive director of community development in Fitchburg, MA reached out to me to discuss the project. We had meetings trying to understand the scope of the project, the goal of the project, and the impact the project could have. We discussed many alternatives and reasons why we needed a custom solution for the vacant buildings downtown. An alternative way was to use Google maps directly, but the information on Google maps or Zillow wasn’t very extensive and the information wasn’t up to date.
We narrowed the problem to this statement: “Since many businesses were closed during the pandemic, it is important to know vacant buildings present in the event new business owners want to utilize the buildings”
Step 2: Storefront Vacancy Data Collection
The next step was to begin collecting vacancy information. Fortunately, the city had some vacancy information that had not been updated in a while.
I was tasked with verifying that the current data provided was up to date and obtaining information that was missing. This step involved conversing with different officials, as well as visiting the actual locations to properly verify the information given.
I spent time taking pictures of actual vacant buildings and converting those images to a format that can be read in the data visualization dashboard
Note: I relied less on technologies here but rather on in-person relationships to build gather domain knowledge. The big takeaway was the importance of improving communication skills as a data scientist because those skills become very invaluable on real projects.
Step 3: Data Storage
The next step was to store both tabular and image data in a location that was safe but easily accessible for important stakeholders to update. I utilized data storage in the cloud and ingested the data to a specific location. The storage piece was very straightforward.
Step 4: Data Cleaning with Tabular and Image Connections
In order to ensure the appropriate images appeared for the data collected, I utilized a process to get the stored image URL location and mapped it to the tabular vacancy information. I also spent some time here getting the contact information of the property owners to be provided in the visualization. Some owners were tough to track down thus making this project an ongoing one.
Step 5: Dashboard Development with Tableau
After spending a sizable amount of time collecting and cleaning data, I began the dashboard development phase building data visualization using Tableau. I utilized Tableau’s data connector to connect with cloud data storage and from there I was able to create a dashboard that included information about the owner, a map view, access to more information on the property, and the image of the particular location. The screenshot of the final dashboard is below. I have grayed some of the vacancy information intentionally for privacy reasons.
Storefront vacancies Dashboard with Map And Images. Image by Author
In conclusion, working on this project was a great experience in building a real world analytic solution and create impact in my community.
Bio: Dotun Opasina is an Associate Data Scientist at Slalom (a modern consulting firm) in Boston. He is passionate about data science, technology, and community impact. He founded SeeksCo to help local business connect better with customers, and DotunData to work on impact projects in communities. Connect with him on Linkedln and see more of his work at www.dotunopasina.com.
- TabPy: Combining Python and Tableau
- How Visualization is Transforming Exploratory Data Analysis
- Data Analysis Using Tableau