Design & Content © 2019 Taoxi Li

Home Searching Made Informed

InHomed

Home-searching made informed

SEP - DEC 2016

Overview

5,250,000 homes were sold in 2015, according to data from the National Association of REALTORS®.  There are existing tools that help potential home buyers understand the real estate market, yet people still struggle to make sense of these data.

 

In this project, we followed a user-centered process to make house-purchasing decisions easier.

MY ROLE
UX Designer
UX Researcher
METHODS
User Journey Mapping
Scenarios
Interviews
Usability Testing
TECHNOLOGIES
Tableau
OpenRefine
Sketch

COLLEAGUES

Francis Estrada

TJ Koines

Handling the data
  • 80, 000+ data points

  • 20 dimensions

  • Northwest Multiple Real Estate Database

Interviews & surveys
  • 5 interviews

  • 13 survey respondents

Storyboard & persona
  • Storyboards and personas created based on research

CHALLENGE

Design a visualization application experience that makes home-searching and price determination learned and comprehensive for potential home buyers in King County, Washington.

"When I first moved to the city, I had no idea where to start looking."

"Don", from our interview

We Studied Existing Works

They did well in terms of

  • Geographic representations of properties currently for sale

  • Detailed information specific to individual properties

  • Comprehensive filtering functionality

But there are limitations...

  • Hard to see trends

  • Little information on neighborhoods

  • Little information available on high-level view

 

Early Design & Findings
We collected and analyzed both quantitative and qualitative data
Findings Summary
  1. The difference between listing and selling price helps buyers decide how much to offer.

  2. The appropriate representation of data matters. For example, average users have difficulty understanding of box whiskers. Bar charts are more intuitive, even if they seem less interesting aesthetically

  3. The need to balance between providing enough information and keeping the cognitive load low is important in order to not overwhelm users.

  4. The interface of Tableau itself is unfamiliar to users. It has a steep learning curve as well, which makes it more difficult for people to use

Inspired by our findings, we made changes to early prototypes and came up with the following visualizations:

Design Changes
  • Embedded the visualizations to a web page, instead of simply showing it on Tableau.

  • Added functionalities to allow users to filter houses by styles (e.g. condos, townhouses etc)

  • Changed the bathroom filter number format into discrete values instead of continuous ones, because it does not make practical sense to have 1.4389 bathrooms. But we did keep numbers with decimals of X.25, X.5 and X.75 because they do have practical meanings[1]. 

  • The box and whiskers chart was removed due to users’ difficulty in understanding this type of visualization.

  • Included data of the past 18 months of houses sold in the King County (May 2015 - December 2016) to allow better analysis of price change trends.

So, how was the result?

We test our work against Ben Schneiderman's principles of visual information seeking mantra[2]. Below is a selection of criteria that we applied.

Overview

The color coding effectively shows which areas are most expensive and which ones are more affordable. 

Zoom

The housing areas visualization allows users to zoom in on zip codes. It thereby gives the user feedback they clicked it, while also maintain a sense of context.

Details on Demand

Once a user has filtered down to a set of comparable homes, they can scroll over the homes to find more in-depth details on the home. 

Filter

Both the housing areas and comparable homes visualizations provide many options for users to filter for data that is most relevant to them. 

Evaluation
  • Partially successful: We provided visualizations that can effectively facilitate home buyers ascertaining the average value of homes in a zip code, the “hotness” or frequency of sales in that area, as well as the trend of the value of homes in that area

  • Drawback: not able to put in the names of areas in zip codes. This makes the areas hard to recognize since most users are not familiar with the boundaries of various zip codes the way they are with the names of neighborhoods

  • Users expressed interest in using the tool to get an overview of the housing market before using a tool like Zillow or Redfin to search for particular houses.

Next Steps
  • Standardize Home Type Definition: The category “Style Code” in our data refers to the type of home e.g. condo, one story, two stories, etc. This category is user-generated which makes it inconsistent at times. We would like to standardize the home types for more clear data.

  • Include Neighborhood Name: As mentioned above we would like to divide our map by neighborhoods instead of zip codes. This would provide more granular data on areas as well as a more understandable interface.

  • More Data: Right now we only have about 18 months of data in our visualization. Ideally, we would have at least 10 years worth of data to give a better idea of housing market trends.

 

References
  1. Rogers, T. K. (2011, February). The 1.75 Bath. Brick Underground. http://www.brickunderground.com/blog/2011/02/the_175_bath_apartmentca. Retrieved October 9, 2016.

  2. Shneiderman, B. (1996, September). The Eyes Have It: A Task by Data Type Taxonomy for Information Visualizations. Proceedings 1996 IEEE Symposium on Visual Languages. doi:10.1109/vl.1996.545307