Tuesday, September 02, 2014



Blog Search



Recent Blog Entries

Visualizing Recovery Funding Requests and Spending

Apr 30

Written by:
Thursday, April 30, 2009  RssIcon

The American Recovery and Reinvestment Act calls for an unprecedented level of spending on a variety of programs, projects, and initiatives across the country. Ensuring the timely delivery of funds to those in need while limiting fraud, waste, and abuse presents both a challenge to those managing the effort, and opportunity to increase the transparency of government spending.

In support of this transparency effort, a number of sites have emerged that are dedicated to tracking stimulus project spending and helping people visualize, analyze and understand how stimulus funding is being allocated including:

  • Recovery.gov – the official U.S. Government sponsored site that tracks stimulus spending
  • Recovery.org – a site sponsored by Onvia that tracks award/disbursement of funding for stimulus projects
  • StimulusWatch.org – an independent site that allows users to vote on the criticality of projects submitted as part of the U.S. Conference of Mayors' report

Although these sites represent great steps in the right direction, none provide a robust geospatial mapping component, which can help make understanding and interpreting the vast array of stimulus data easier. Maps help people visualize data and absorb information more quickly and intuitively than tabular formats. Within the context of the recovery effort, leveraging geospatial visualization and analysis tools would enable people to see, understand, evaluate, and provide feedback on stimulus spending more easily.

Visualizing stimulus projects from StimulusWatch.org

To illustrate some of the ways recovery spending can be visualized, we leveraged data from StimulusWatch.org to create a series of geospatial mashups using Microsoft's Virtual Earth. It should be noted that the techniques used to create the maps in this post can be applied with other mapping tools like Google Maps.

At the time of download, StimulusWatch.org had collected a total of 676,807 votes measuring the criticality of 18,750 proposed projects in 785 cities across the country. The diverse nature of this dataset and the availability of city and state information made it a prime candidate to display using heat mapping. While there are a number of methods for visualizing geospatial data, heat maps can be an ideal method for visualizing data where there are too many points to show individually on a map, or where doing so would not aid analysis.

The map below depicts the density, or number of proposed stimulus projects, submitted by each city in the U.S. Conference of Mayor's report. From the map below, we can see the Miami, LA, and Chicago areas proposed the most projects.

Heat map of the number of proposed stimulus projects by city

Data source: StimulusWatch.org

In addition to plotting the density of projects on StimulusWatch.org, we also mapped the aggregate project value to show areas with the highest proposed levels of spending. Based on the map below, we can see that projects in Puerto Rico have the highest cost, followed by projects in the Miami, LA and San Francisco areas.

Heat Map – Dollar value of project requests aggregated by city

Data source: StimulusWatch.org


In addition to providing data on project location and cost, StimulusWatch.org also allows users to rate and comment on stimulus projects. The voting and commenting system implemented by StimulusWatch.org allows users to rate the merit of proposed projects and helps establish a general consensus of relative project value.

The map below illustrates the areas of the country where projects received the most attention, with projects in Southern California, Texas and Gulf Coast, as well as Ohio/Kentucky region attracting the most votes. This map helps answer the question, "Where are the projects that people are most interested in?"

Heat map of the concentration of project votes aggregated by city

Data source: StimulusWatch.org

By overlaying aggregate vote ratio data on a map, areas where users feel strongly about projects can be illuminated. The map below highlights areas where projects were voted to be the "least critical" by users. Project criticality was measured using StimulusWatch.org's "vote ratio" calculations, which were aggregated by city, and then clustered to generate a heat map. From the map below, we can see that proposed projects in Puerto Rico (not shown), Mississippi, Texas and California received a relatively high level criticism from StimulsWatch.org users.

Heat map of projects voted least critical aggregated by city

Data source: StimulusWatch.org

Likewise, projects in Illinois, Kentucky, and southern California received the highest positive vote ratio. The map below helps illustrate where there is a general consensus about the merit and valuable nature of proposed projects.

Heat map of projects voted most critical projects aggregated by city

Data source: StimulusWatch.org

In addition to using heat maps to visualize StimulusWatch.org's data, thematic maps, or maps illustrating data within the context of familiar boundaries such as states, counties, and zip codes can also be used.

The following map illustrates both the aggregate dollar value of proposed stimulus projects, and the number of people in the workforce who are unemployed within each state (not unemployment rate as a percentage). Aggregate project dollar value is represented by color (purple = highest, blue = lowest), and the number of unemployed is represented by the height, or altitude of each state.

Dollar Value of StimulusWatch.org funding requests vs. # of Unemployed by state

Data source: StimulusWatch.org, Bureau of Labor Statistics

From an accountability perspective, it could be reasonable to expect there to be a correlation between the level of unemployment and allocation of stimulus spending. A quick glance at the map above illustrates this to be the case for the most part – California has both the highest level of unemployment, and the greatest aggregate project dollar value. Likewise, Texas and Florida have corresponding levels of unemployment and funding requests, while New York has disproportionally low level of funding requests for its level of unemployment, and Puerto Rico has a disproportionally high level of funding requests for its overall unemployment.

Moving Forward

Recovery.gov is still in its infancy and generally does not provide users with a deep-dive of stimulus spending data. Additionally, Recovery.gov does not currently provide any geospatial visualization and analysis tools. The good news is that Recovery.gov is seeking input from the community on ways to increase the transparency and effectiveness of the site, and is currently soliciting input via http://www.thenationaldialogue.org from April 27th through May 3rd.

There have already been a number of ideas posted on the thenationaldialog.org that suggest adding geospatial visualization and analysis capabilities, however, I would advocate that any potential solution also consider:

  • Providing Recovery.gov data through Application Programming Interfaces (APIs):  Consider providing recovery.gov data for public consumption through various APIs such as GeoRSS/Web Services.  Enabling this functionality would be extremely cost effective since it would allow the community to develop additional functionality independently of activities funded by recovery.gov. 
  • Integrating additional data sources with Recovery.gov data:   Consider adding other relevant geospatial data layers beyond stimulus specific data, such as the unemployment rate, so that users can visualize relationships between funding requests/awards and local economic conditions. 
  • Enabling Citizens to provide feedback on Stimulus spending: Consider implementing a voting system, similar to that of Stimuluswatch.org, that allows users to rate and comment on spending. 

Solution Architecture

  • ASP.NET/SQL Server 2008/Microsoft Virtual Earth
  • Dynamic heat maps were created based on the approach described in Johannes Beck's blog, with modifications to leverage geospatial indexing/querying capabilities of SQL Server 2008.
  • Thematic maps were created using VE Shape Layers/VE Shapes, with boundary files as the polygon sources, passed via JSON web service


Data Sources

  • Information on proposed stimulus projects was downloaded from StimulusWatch.org. StimulusWatch.org's project vote data was retrieved using a custom crawler/regular expression parser. Update: StimulusWatch.org has been kind enough to provide a JSON API that can be used for future updates (Thanks Pete).
  • Unemployment Data was retrieved from the U.S. Bureau of Labor Statistics
  • City latitude/longitudes were retrieved from the U.S. Census Bureau and other geocoding resources. Generalized cartographic boundary files were retrieved from the U.S. Census Bureau.


NOTE: The observations, comments, and statements in this post are intended simply to convey interpretation of data retrieved from StimulusWatch.org and do not represent Strategi Consulting's views or position on stimulus project spending.


Your name:
Your website:
Security Code
Enter the code shown above in the box below
Add Comment   Cancel 
Copyright © 2013 Strategi Consulting LLC