American Red Cross of Chicago
Contents |
Team Members
Aaron Wolf, Alejandro Morales Gallardo, Anne R., Antony Thomas, Ashley Asaju, Dan Morgan, Gabriel Gaster, German Pic, Jim Murphy, Jo Strang, Marshall Smith, Mike Stringer, Neus Herranz, Paul B., Paul Davis, Rob Lancaster, Sheri Gilley, Silvia Eiden, Young-Jin Kim
Source(s): American Red Cross of Chicago
Approach
The key overall question that we set out to address was strategic: where should the Red Cross allocate resources to more effectively prevent disasters?
Because they account for the bulk of Red Cross activity, we sought to understand domestic fires in particular.
To do this, we compared the density of disasters geographically, and ran correlation tests for a number of variables, including demography, population density, among others.
Source(s): American Red Cross of Chicago
Findings, Results, Assorted Graphs and Such
Fires happen more in certain places
The first thing we noticed was the strikingly unequal distribution of disasters. While disasters do happen all over Chicago-land, outside of a small number of zip-codes, they are rare events. While it's good that the Red Cross responds to these disasters, it would seem that prevention efforts are best allocated to areas where disasters are less rare.
Neighborhoods such as Austin, Back of the Yards, Englewood, West Englwood, Roseland, and Harvey have significantly higher concentrations of DRs.
A heat map showing total DRs per zip-code immediately shows that the concentration of DRs [INSERT GRAPH]. However, normalizing for number of households per zip-code shows the concentration is actually more acute [INSERT GRAPH].
Correlations
DRs per capita are strongly correlated to: per capita income, race, and population, although not population density. We found that DRs per square mile are strongly correlated to population per square mile.
These graphs show the distribution of average incomes by zipcode, overlayed against the distribution of average incomes among high DR-zip-codes. INSERT GRAPH LINK HERE
Population density is not correlated to DRs, although overall population in a zip-code is.
Making the data accessible
With the following google fusion tables, the Red Cross can now explore the geographic spread of their data according to a variety of cuts.
Here is the data of disaster responses (normalized by number of houses) by zipcode, in a Google Fusion Doc. And here is a link to the map.
We see from this map that a small number of areas have a high number of fires, while the majority of areas have very few fires.
Consistent with the above, if we order the zip codes by the number of disaster responses (decreasing) and plot the cumulative percentage of DRs as a function of the number of zip codes included, we get the graph below. From this we can see that 80% of disaster responses occur in 20% of the zip codes. Furthermore, 60% of the the DRs occur in 10% (about 22) of the zip codes. Prevention efforts can be concentrated in these zip codes for best effect.
Perhaps not surprisingly, there is a rough (reverse) correlation between income and the number of disaster responses. We see that fewer DRs are likely to occur at higher income levels.
Similarly, we see a rough correlation between the median age of structures in a (zip code) area and the number of DRs.
The median structure age and income also show correlation. Identifying causality is trickier, but we see that impoverished regions generally have older structures and it is possible that the trigger event for many fires is the age of the structure. Concentrating prevention efforts in low income areas with older structures may have the biggest benefit.
Source(s): American Red Cross of Chicago
Open Questions / Suggestions / Gripes
Ideas
- Aggregate by "responder region" instead of zip
- Correlation between income per capita / DR per capita / age of buildings / race / age
- Heat maps by zip code for above metrics
- Scour the data for "interesting cases," and see if Red Cross has any insight into those cases.
- Examine response time for high-density DR zip-codes. American Red Cross could advocate for faster response times in high-density areas to control the damage of fires in areas where they are less rare.
Source(s): American Red Cross of Chicago
Questions
- Should Red Cross concentrate on low income areas, or is it individual low income homes that are more likely to have a fire? -- This would require data about household income of disaster.
- What data can the Red Cross start tracking that will help with prevention / mitigation?
Source(s): American Red Cross of Chicago
Data, Links, Other Info We Rounded up
Chicago Red Cross data:
- Incidents July 2010-December 2011.csv -- all disaster response data
- 20120224 Cook County Donations by Zip revised.csv -- donation by zip code for a limited time period (July 2011-December 2011)
From City of Chicago data portal:
- 311_Service_Requests_-_Vacant_and_Abandoned_Buildings_Reported.csv
- Fire_Stations.csv
Census data from http://www.factfinder.gov
- 2000income/DEC_00_SF3_P082_with_ann.csv
- 2010housing/DEC_10_SF1_H1_with_ann.csv
- 2010population/DEC_10_SF2_SF2DP1_with_ann.csv
- 2010race/DEC_10_SF1_QTP3_with_ann.csv
- tl_2010_17_zcta510.* -- geographic boundaries for zip codes in IL
- illinois_zips.kml (converted to KML from shapefile using ogr2ogr)
From US Fire Administration at http://www.usfa.fema.gov/downloads/xls/
- qr_summary_2007.csv -- US fire fatalities in 2007
- qr_summary_2008.csv
- qr_summary_2009.csv
- qr_summary_2011.csv
API for getting census blocks
Incidents data with 'official' google address, longitude, latitude data
Conflicting addresses:
Using "3012 S Spaulding Ave, Chicago, IL" for "3012 S. SPAULDING 46306 GARY"
Using "1108 East 91st Street, Chicago, IL 41.729711 -87.596947" for "1108 EAST 91ST ST, SCHILLER PARK, 60176"
as reasonable matches. description.txt (END)
• Back to Chicago DataDive 2012 main page
Source(s): American Red Cross of Chicago
