Skip to content

Households per Census block

9 messages · Keith S Weintraub, Anthony Damico, Zack Almquist +1 more

Keith S Weintraub
# 
Folks,

I am using the UScensus2010 package and I am trying to figure out the number of households per census block.

There are a number of possible data downloads in the package but apparently I am not smart enough to figure out which data-set is appropriate and what functions to use.

Any help or pointers or links would be greatly appreciated.

Thanks for your time,
Best,
KW
Anthony Damico
# 
hi, ccing the package maintainer.  one alternative is to pull the HU100
variable directly from the census bureau's summary files: that variable
starts at position 328 and ends at 336.  just modify this loop and you'll
get a table with one-record-per-census-block in every state.

https://github.com/davidbrae/swmap/blob/master/how%20to%20map%20the%20consumer%20expenditure%20survey.R#L104

(1) line 134 change the very last -9 to 9
(2) line 137 between "pop100" and "intptlat" add an "hu100"


summary file docs-

http://www.census.gov/prod/cen2010/doc/sf1.pdf#page=18
On Mon, Aug 3, 2015 at 11:55 AM, Keith S Weintraub <kw1958 at gmail.com> wrote:

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Zack Almquist
    

    

      
      #
    

  


  

      
Hi Anthony and Keith Weintraub,

Here is a way to do what you are asking using the UScensus2010 packages:

## latest version of the package, not yet on CRAN
install.packages("UScensus2010", repos="http://R-Forge.R-project.org")
library(UScensus2010)
install.blk()
library(UScensus2010blk)
### You will want the H0010001 variable (see help(alabama.blk10))
### Other variables are also available
### You can use the new api function in UScensus2010 to get arbitrary
variables from SF1 and acs

data(states.names)
head(states.names)
state.blk.housing<-vector("list",length(states.names))
## notice this could be greatly spead up using the library(parallel)
## with mclapply
## This will be somewhat slow b/c of so much spatial data
for(i in 1:length(states.names)){
data(list=paste(states.names[i],"blk10",sep="."))
temp<-get(paste(states.names[i],"blk10",sep="."))
#unique b/c more shapefiles than fips
state.blk.housing[[i]]<-unique(temp at data[,c("fips","H0010001")])
print(i)
rm(paste(states.names,"blk10",sep="."))
}

###########
# alternatively Using the US Census API function in the new UScensus2010
package
###########

## Get all states fips code
data(countyfips)
state.fips<-unique(substr(countyfips$fips,1,2))
head(state.fips)
length(state.fips) ## will be 51=50 (states)+ 1(DC)
## You will need a census key
key<-"YOUR KEY HERE"
housing<-CensusAPI2010(c("H0010001"), state.fips=state.fips, level =
c("block"), key, summaryfile = c("sf1"))

Best,

-- Zack
---------------------------------------------------------
Zack W.  Almquist
Assistant Professor
Department of Sociology and School of Statistics
Affiliate, Minnesota Population Center
University of Minnesota

        
On Mon, Aug 3, 2015 at 12:43 PM, Anthony Damico <ajdamico at gmail.com> wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Keith S Weintraub
    

    

      
      #
    

  


  

      
I had to download a bunch of stuff but I got it mostly working.

Unfortunately using the alternative method I get the following:

            

      
      
      
    

        
Error in file(con, "r") : cannot open the connection
In addition: Warning message:
In file(con, "r") : cannot open: HTTP status was '400 Bad Request?

I have a feeling that this is not a problem with the API.

Thanks for your help,
KW

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Zack Almquist
    

    

      
      #
    

  


  

      
Hi Keith,

        
On Tue, Aug 4, 2015 at 12:43 PM, Keith S Weintraub <kw1958 at gmail.com> wrote:

        

            

      
      
      
    

        
Sorry that is my bad, I didn't verify the variable name at (
http://api.census.gov/data/2010/sf1/variables.html). This seems to work for
me, as a quick test:

housing<-CensusAPI2010(variables="H0060001", state.fips="06", level =
c("tract"), key, summaryfile = c("sf1"))

So the larger example:

 ## Get all states fips code
data(countyfips)
state.fips<-unique(substr(countyfips$fips,1,2))
head(state.fips)
length(state.fips) ## will be 51=50 (states)+ 1(DC)
## You will need a census key
key<-"YOUR KEY HERE"
housing<-CensusAPI2010(c("H0060001"), state.fips=state.fips, level =
c("block"), key, summaryfile = c("sf1"))

should work just fine.

Best,

-- Zack
---------------------------------------------------------
Zack W.  Almquist
Assistant Professor
Department of Sociology and School of Statistics
Affiliate, Minnesota Population Center
University of Minnesota

            

      
      
      
    

        

      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Zack Almquist
    

    

      
      #
    

  


  

      
P.S. The US census has different "populations" (or worlds) so make sure the
housing variable you use is accessing the correct "world."

Best,

-- Zack
---------------------------------------------------------
Zack W.  Almquist
Assistant Professor
Department of Sociology and School of Statistics
Affiliate, Minnesota Population Center
University of Minnesota

        
On Tue, Aug 4, 2015 at 2:12 PM, Zack Almquist <almquist at umn.edu> wrote:

        

            

      
      
      
    

        

      
      
    

        

      
      
    

            

      
      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Keith S Weintraub
    

    

      
      #
    

  


  

      
Can you give me a for-instance of ?populations?? Is there a table or chart or list or?

Also I guess that I should leave my R session on for as long as possible as ?install.blk? takes a really long time to re-upload if that is what it is doing.

Does install.blk go to the source every time?

Thanks again.
---
KW

            

      
      
      
    

        

      
      
    

            

      
      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Zack Almquist
    

    

      
      #
    

  


  

      
Hi Keith,

I would only use install.blk() once. Then just load the
library(UScensus2010blk) like normal on your machine (this should be
relatively fast), redownloading and re-installing each time will be very
expensive (both on download and time).

The SF1 file manual produced by the US Census goes over all the details of
the various variables (http://www.census.gov/prod/cen2010/doc/sf1.pdf) in
glorious detail; however a lot of the issues arise depending on if you are
interested in individual counts versus household counts and assumptions
thereof.

Best,

-- Zack
---------------------------------------------------------
Zack W.  Almquist
Assistant Professor
Department of Sociology and School of Statistics
Affiliate, Minnesota Population Center
University of Minnesota

        
On Tue, Aug 4, 2015 at 2:31 PM, Keith S Weintraub <kw1958 at gmail.com> wrote:

        

            

      
      
      
    

        

      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      David Winsemius
    

    

      
      #
    

  


  

      
On Aug 4, 2015, at 12:31 PM, Keith S Weintraub wrote:

        

            

      
      
      
    

        
I have no experience with that package, but the help page says:
======:
Warning:

This is an extremely large file (around 2 gigs) and should only be installed if you have a very good connection. Also it is worth noting that for all systems the install is from source and can take quite a bit of time to install.
======:

It also says you need to provide an argument to the function: one of "osx", "linux" or "windows".

I would have guessed that you only need to run it once, and after reading its source continue to think that there should have been a package installed, which would then persist and not need to be repeatedly installed.