Introduction

Housing is one of many basic needs that has been effected by the COVID-19 pandemic. As a response to tenants’ inability to pay rent due to statewide lockdowns, the first state eviction moratorium of commercial and residential property was announced on March 20, 2020 by New York Governor Andrew Cuomo. Both its effectiveness and relationship to the overall housing crisis can be analyzed, in part, by using these graphs.

Materials

These graphs demonstrate the decreasing rate of eviction from 2017 to 2021 utilizing a dataset from NYC Open Data. I found a dataset relevant to the eviction rates in NYC containing both quantitative and categorical dimensions. The dataset followed count over time with more than 66,000 entries. A CSV file was available to download, which served as my data source for the free online visualization software Tableau Public that allows users to publish and explore user-created content. I registered for an account with the platform and watched How To Videos to help navigate the software’s features, which encourages contextualization through dashboards and interactive visualizations.

Process

After I connected to the dataset, the visible fields helped me explore ways to filter and highlight data. My goal was to analyze count over time which is best represented through a line graph. I referenced visualizations on the same topic in multiple ways to observe how datasets can be manipulated to enhance meaning. Through this research, I was eager to see some examples that were not so elegant in transforming datasets across multiple charts and graphs. Having good and bad examples helped in making design decisions for this visualization.

My first sheet below contains a line graph. It has continuous values on both axes, as illustrated with a green pill. To shape the data, I applied filters and separated discrete values. When I added Years to the graph’s columns, I noticed that the dataset included some entries under the year 2070. I interpreted this as an error since neither multiple entries in this year were present, nor projections were mentioned in the dataset description.

Then I created multiple lines to represent the 5 boroughs. When choosing a color scheme, I went with the choice listed as visible for people who are color-blind. It seemed like the best choice for accessibility. Colors were assigned randomly, but I changed them to all be around the same level of boldness as to not create emphasis on any specific borough.

I wanted to edit out superfluous features and address titling. In the formatting tab, I removed grid lines from the graph. Then I edited the axes to be human readable and applicable to the data. Initially, the axes titles were automatically populated with the names illustrated below in the Columns and Rows. The sheet name was edited to NYC Evictions Executed.

To emphasize a historically significant event that may effect the data, I chose to insert a reference line. In the Analytics tab, I found the option to do so. There, I found a way to add a fixed line on the date March 20, 2020 representing the first statewide eviction moratorium. I chose a dotted line to imply that the rates would move through the event.

After finishing this sheet, I realized that the 5 lines in the graph merged very closely together making it difficult to see what happens after the reference line I created. Since that represents a historically significant event, I thought it was important to expand the period between 2020 and 2021 into another graph below.

Here I chose to adjust the Year in the Columns to Month in order to more clearly represent the spread of data. Then I changed the format from lines to open circle. This way overlapping data points and single entries missing values are present. Values of zero are not included in the dataset, so these absences seem to represent no data.

I kept the color key consistent, so when creating a dashboard the relationship would be clear. I also chose to, again, delete the grid lines and title the sheet, this time right justified for flow.

Finally I added a caption instead of keeping all of the months labeled below each row. It is still clear and when hovered over, can be read. The caption lists the year as well.

Results

The Dashboard representation uses space economically while communicating what each datapoint references. Its interactive components allow viewers to highlight a specific borough in both graphs simultaneously by clicking on the borough names within the key. This way, when stacked values become less visible, it is easier to read an exact value by scrolling over an open circle or a line.

I strategically chose where to add text in the dashboard tiles. The main title, axes, secondary title and caption simulate a diagonal similar to that of the graphs, while the color key is nestled in the negative space to the left. The goal was to create a flow, and make it clear how to interact with the visualization.

Reflection

My priority was to make a concise visualization while also raising some questions. While I expected declining eviction rates due to legislative action, I know that these cannot be a complete picture of what is happening during the current housing crisis. How can these graphs be used in journalism, non-profit aid organizations or by politicians? Each perspective would lead to a different usage and interpretation of the dataset. Perhaps someone is interested in portraying the response to COVID-19 as effective, and these graphs are used, in part, to bolster that claim. My hope is that these can be an initial step to a larger investigation into the ways NYC eviction rates are declining. More vizzes related to NYC housing datasets can be used to create further context and flush out what the graphs will be used for in future iteration.