10 Years of American Health
Data Visualization Project

Visualizing trends in Google search data to better understand American lifestyle

Project Brief
For this Project, I was interested in comparing trends in American Google search behavior to health (specifically obesity). To do this, I used Processing to create an interactive visualization that explores the last 10 years of American health and google search data. I incorporated two major data sets into this visualization: Google search terms across American cities (provided by Google Trends), and American obesity by state over the past decade (provided by Trust for America's Health). My goal for this project was to design an engaging and interactive visualization that helps viewers to reach new understandings and make discoveries about patterns in American behavior.
Date: Winter 2015
Role: Independent Project
Design Process
My initial concept sketches were very basic: Blue circles indicated cities most interested in "fitness" (based on Google search data), and red circles indicated the most obese states in the country. The trendline at the top indicates the overall interest in fitness throughout America at a specific time (month, year). Users could hover across the trendline to view the data at different points in time. This would allow users to easily view and compare changes in the data over time.
As I began sketching in code, I started to lay out the page according to my sketches. Here I used data from a single year (2004) just to begin figuring out how the information might be communicated visually.
In order to refine the concept further, I added date and city information on hover. I also reduced the opacity of city circles to improve the visual aesthetic and allow overlaps to be more clear.

As I refined the functionality of my prototype, challenges began to surface. For example, it became clear that accurately mapping location data to my map was difficult. I took several iterations to address this problem.
Darker background tones made the important information pop.

I added a toggle button to show users what search term data is being shown (in this case "fitness").
I moved the timeline to the bottom of the visualization to improve the information hierarchy of the visualization. The most important information is now the first thing that a viewer will look at.

Changed the granularity of the timeline (years) to match the information shown on the map. The selected year is highlighted on the timeline to help communicate this organization and make yearly patterns clear.

A ring around the hovered city further emphasizes the selected city.

Added "workout" search data and included a button to control the visualization of that information.
Changed the title to clarify and add emphasis.

Distinguished individual states with outlines, and communicate obesity by background color. This structure solved the location mapping problem, and allowed me to clearly show obesity data for every state. The most obese states are now listed as opposed to being identified by circles. This helps distinguish search data from obesity data.

Added "weight loss" search data and toggle button

Added question buttons to show additional information on hover

Changed the color balance to give the search data more emphasis.

Included a key to make the obesity information more clear.

Removed the top 10 most obese states list because that is not the most critical information; the general pattern of obesity compared to search trends is the focus rather than the specific ranking of obesity.

Consolidated additional information into a single question button to reduce clutter.

Final Iteration
Simplified the title to make it more clear.

Changed the toggle buttons to check boxes to make it clear that they are clickable.

Removed workout data because there was no clear pattern associated with it. This reduced clutter and made the important patterns most evident.

Reflection & Analysis
In working through this project, I honed my skills in information visualization, and gained valuable experience with rapid prototyping and iteration in software. This experience helped me to feel comfortable using code to quickly prototype and test visuals, interactions, and functionality.

As for findings related specifically to this project, I made multiple discoveries about American search interest related to location health. The two most evident trends were:

1) Nationwide search interest in terms related to health/fitness peaks at the beginning of January and plummets every winter. This trend is extremely consistent throughout the past decade, and is undoubtedly linked to new year's resolutions.

2) The cities most interested in terms related to "fitness" typically are located near the coasts and almost never overlap with the most obese states. However, the cities most interested in "weight loss" overlap heavily with the most obese states. This means that while "fitness" and "weight loss" have very similar meanings, they have different uses. I hypothesize that for those who are very over weight, weightloss is the largest concern. But, for those who are normal weight, fitness is the next goal beyond weightloss.

back to projects