Like many people my age, I rent. I’m fascinated by the rental and housing markets and am always interested to see how markets are developing in different cities and regions across the US.

An article on Zumper, an apartment rental website, caught my eye last week during lunch. It was called “Mapping DC Rent Prices This Summer”, and as a DC resident and data enthusiast, I naturally dove right in. The article analyzed rent prices for each of the major neighborhoods in DC, drawing on data from their June 2016 “Zumper National Rent Report”.

I uploaded a data set based on their 2016 report and built the dashboard below that explores short- and long-term movement in rental prices for each of the top 50 metro areas in the US. You can check it out here, or keep reading below to learn more about how I put it together.

After a bit of digging on the Zumper site I was able to locate the full dataset. I copied the table, opened up Desktop and began the data import process. Since the data was already in a tabular format, I decided to use the Clipboard data import option. I realized that there were some things that needed cleaning up before I could conduct my analysis, so I opened the data preparation interface.

Because of how the table was structured in HTML, MicroStrategy was reading the 1 Bedroom Rent Price and 2 Bedroom Rent Price fields as the only column headers. I parsed the data to get around this, a feature that allows you to control how MicroStrategy treats imported data. In this case, I told the program to ignore the top row—resulting in the correct column headers.

Since the top row was how the original table differentiated the metrics for 1 and 2 bedroom apartments, I had to make each of the column names unique—by adding “1 BR” and “2 BR” to their respective metrics. I also went ahead and changed the name of the Pos. attribute to Rank.

My first analysis explored potential relationships between short and long-term changes in rental prices. I selected a bubble chart from the visualization gallery and dragged the M/M% and Y/Y% metrics for 1 BD apartments into the drop zone and City into the break by field.

This gave me the following four-quadrant visualization. The top right represents cities that have seen both long- and short-term growth, while the bottom right shows short-term growth and long-term decline. To make these four quadrants stand out, I decided to add two reference lines, one for each metric at 0%.

I then replicated this visualization and swapped out the 1 BD metrics for their 2 BD equivalents, and added a simple grid with City and Rank. I hooked this grid up to act as a filter on the two visualizations, allowing me to highlight the individual data points in each visualization for the appropriate city.

After looking at this analysis, I realized that coloring by State was distracting, as there were too many different attribute elements being visualized. In general, it's best to keep the number of colors on a single visualization to no more than 6, and I had almost 30. So I decided to bring in another dataset that I had pulled together for a different dashboard that groups the states into regions as defined by the US Census Bureau. I was then able to link these datasets on the shared State attribute, and color by region—resulting in the final dashboard.