Project Background

Home prices are growing crazily

Housing and rental prices have always been subject to the whims of the market, fluctuating over time, rising and falling based on supply and demand. Many companies have spent untold dollars monitoring and logging this data to give researchers and comment people the ability to view the current market within the context of data spanning decades to establish trends and compare results.

Our project and final visualization explores the historical housing and rental market fluctuations over the last 20 years across the United States. The goal was to visually represent how different areas in our country have grown or diminished, and give the user the ability to manipulate that data through filters, such as property type and features.

Course
Information Visualization (HCDE511)
‍

PracticesData Visualization, Usability Studies

Methods & Tools
Tableau, Sketch, Principle

Role
‍UX Designer & Researcher

Date
‍October-December, 2017

Teammates
Will Bishop (Data Scientist),
Mike Knauer (UX Designer & Researcher)

Here is the link of our work.

This project displayed data from Zillow Research and is created with the visualization software Tableau.

Process

Discover, define, design, and deliver

When we started this project, we elected to use the basic User-Centered Design process of Discover, Define, Design, Deliver. This process allowed us to research and explore our chosen space and better understand what our users needed and the best options for delivering a solution. As we moved forward we were able to iterate on our design and validate the direction of our project. Overall, this process was a great means to ensuring that our project delivered a solid solution.

User Research

What are home buyers looking for?

To get greater insight into our users, we sent a survey to more than 60 people to get an understanding of what people who own a home - or are looking for a home - would find interesting as they explore a visualization. This data validated our initial hypothesis that users were interested in viewing historical housing trends by year and property type, as well as understanding how a neighborhood and city has changed over the last 20 years.

Finding #1: Home price and location are the most important factors when users consider renting/ buying a house.

Finding #2: Most users would like to explore the home price trend by neighborhood and years, as well as house size and type.

Data Profile

Expansive Zillow Dataset

The housing price data we used came from two indices with Zillow Research: the Zillow Home Value Index for house prices, and the Zillow Rent Index for rentals. Both of these datasets are intended to list average home and rental prices for all homes in an area, not just those currently listed. The Zillow data covered the entire United States and was separated in multiple levels of geographic area, including state, metro area, city, zip code, and neighborhood. While this data may have been more expansive than we required, having the flexibility could prove valuable. The data was also broken down by type of home: single-family vs. condo vs. multi-family, as well as the size of home (1-bedroom, 2-bedroom, etc.).

Ideation

Explore the visualization

Once we better understood our idea and explored our datasets, we began additional ideation for our vision. We quickly settled on the basic components and layout of our visualization through several quick rounds of sketches. We explored interactions around our map and how we could display states, metro areas, and cities that would allow users to click on different geographic areas and display the home value and rental price data. We also reviewed patterns that expressed different ways users could filter the data and modify the map and associated information.

Solution

An interactive map + a dynamic line chart

After exploring various visualization ideas, we decided to deliver an interactive chart that allows users to explore a map, selecting a specific state or city/metro area and gain historical insight into the housing and rental prices. Users should be able to explore the data and see a chart that displays historical trends, with the ability to filter the data to only display one, or all, of the primary categories.

Usability Studies

Simplification needed. Less is more.

With a more cohesive visualization and interaction set, we created an initial prototype in Tableau and conducted a usability study to validate our initial findings and gather additional feedback from our users. For our study, we used a working prototype of our visualization that was created within Tableau and displayed as a dashboard for users to navigate. At the start of each session, we asked two introductory questions about the visualization, followed by asking them to complete three tasks, which entailed interacting with the map and filtering the data. After completing all three tasks, we asked the participants five follow-up questions regarding their overall impressions.

Task #1‍
You are looking to find out historical information for 3 bedroom home prices in Seattle, WA. How would you modify the visualization to view this information?

Task #2
As you look at the data, you decide to look at the house prices for Seattle from 1996 to 1998. How would you do this?
Follow up: How do you feel about this feature? Is it important?

Task #3
Having explored Seattle, how would you look at historical rental prices in Chicago, Illinois?

Findings & Suggestions

Our usability study showed that our participants had an overall positive first impression of the visualization, despite functional issues related to a low-fidelity prototype. Our primary takeaway from this experience was the need for simplification. Our original goal was to provide an in-depth experience whereby users could examine data in a specific area and then view historical trends. As our ambitions grew, so too did the complexity of the visualization.

In addition, our users had a great deal of feedback around the map. Users had preconceived notions of how the map should navigate and we were often compared to more mature mapping options, such as Google Maps or Zillow Map Search. After this study, we felt the best solution would be to limit selection options just states, which also decreases the potential for users to become lost in the map due to excessive zooming or panning.

Iteration

Three Major Improvements

#1. Limited the Scope
As a result of our study, we limited the scope of our visualization and implemented several minor enhancements. This included limiting the map to only allow users to select states, removing the ability to transition the map to view counties and cities. We also made cosmetic changes to the map, including removing all the other countries using Tableau’s “opacity” parameter, and making Alaska and Hawaii separate maps off to the side in order to zoom in on the contiguous 48 states.

#2. Get Connected
Another important change was to correct minor issues linking the map to the line chart using dashboard actions in Tableau. With the new experience, when a user hovers over a state on the map, that state’s data is highlighted on the line chart; when a user selects a group of states, the line chart filters to show only those states.

#3. Simplified the Filters
We also reduced the number of filters on the “State map and time series” dashboard allowing users to filter by date range, home size (number of bedrooms), home type, and index type (home value, rent, or rent per square foot). When users explore those filters their changes will apply to both the map and the line chart.

Final Product

Interactive Home Price Visualization

Based on our explorations and study, we felt very confident in our direction going into the development of our final visualization, which can be viewed here , as well as screenshots below.

Reflection

What we did well

As a group we feel that we were successful in developing and implementing a visualization that allows user to view and explore housing and rental prices across the United States. We provided an overview, filtering methods, and details on demand, and we appropriately used hue to encode nominal data and value for quantitative data. We avoided chart junk and provided clear legends, titles, and disclaimers.

While the scope of the project changed from the start of our project, the final visualization adheres to our original vision, incorporating many of the features we desired. While we were unable to add the granularity in the map we desired, our testing identified this as a cartographic limitation in Tableau, so we understand that users would not have gotten the benefit of zooming in to view counties and cities as we had intended. Since this was uncovered during user testing, we feel that our process was able to define that issue and give us enough time to address it properly, which would label this change a success.

What could be done differently

An area where we feel we were not entirely successful was alerting users that they had the ability to compare two areas on our map. While the functionality is available through the use of the Control key to select multiple states or by brushing over a group of states, we believe that this could be more straightforward with the addition of a short message or notification to the user. Additional exploration was deprioritized due to time constraints. As a future study topic, this could be a quick and easy include that users could find very interesting.