How do I find other zipcodes that touch a particular zipcode?

Question

I want to create a matrix for around 200 zipcodes and neighboring zipcodes that touch those zipcodes. The matrix would be 200*200 with 1's for the cells in which the two zipcodes touch and 0 when they are not neighboring zipcodes.

How could I create or get such a matrix? Thank you very much.

Best,

Corey

There's a zip code shapefile here, if that helps. – Aultman

From what information would you like to construct this? Do you have a shape file? Perhaps coordinates? – Trihydric

I have access to some of the shapefiles as used by jbaums. But not to all the states that I wanted. I think I could manually input the coordinates of the zip codes if it came down to it. How do you think I should proceed? I wanted the information for 9 states and I was able to get the shapefile for 4 states but not the remaining 5 as I mention below. Please let me know how I can use the lat long information to create this matrix. Thank you very much. – Corey

@danny117 - you're asking a different question in your bounty description, but I think the edits to my post might help. – Aultman

Answer 1

If you have access to a shapefile, this is relatively straightforward with the help of the spdep package.

Here's a standalone example using Californian zip code data (~3.5MB download):

# load libraries
library(rgdal)
library(spdep)

# download, unzip and import shapefile
download.file('http://geocommons.com/overlays/305142.zip', {f<-tempfile()})
unzip(f, exdir=tempdir())
shp <- readOGR(tempdir(), 'tigerline_shapefile_2010_2010_state_california_2010_census_5-digit_zip_code_tabulation_area_zcta5_state-based')

# identify neighbours for each poly
nbs <- setNames(poly2nb(shp), shp$ZCTA5CE10)

# convert to a binary neighbour matrix
nbs.mat <- nb2mat(nbs, zero.policy=TRUE, style='B')

# see?rgeos::gTouches for an alternative to the above steps

# assign zip codes as dimension names
dimnames(nbs.mat) <- list(shp$ZCTA5CE10, shp$ZCTA5CE10)

For our dataset, this returns a 1769 x 1769 matrix indicating which zip codes are neighbours. The first 10 rows and 10 columns look like this:

nbs.mat[1:10, 1:10]

##       94601 94501 94560 94587 94580 94514 94703 95601 95669 95901
## 94601     0     1     0     0     0     0     0     0     0     0
## 94501     1     0     0     0     0     0     0     0     0     0
## 94560     0     0     0     0     0     0     0     0     0     0
## 94587     0     0     0     0     0     0     0     0     0     0
## 94580     0     0     0     0     0     0     0     0     0     0
## 94514     0     0     0     0     0     0     0     0     0     0
## 94703     0     0     0     0     0     0     0     0     0     0
## 95601     0     0     0     0     0     0     0     0     0     0
## 95669     0     0     0     0     0     0     0     0     0     0
## 95901     0     0     0     0     0     0     0     0     0     0

Optionally, if you want a two-column matrix giving neighbouring pairs of zip codes (i.e., zip code in col 1, and neighbouring zip code in col 2), you can use the following.

nbs.list <- sapply(row.names(nbs.mat), function(x) names(which(nbs.mat[x, ] == 1)))

nbs.pairs <- data.frame(zipcode=rep(names(nbs.list), sapply(nbs.list, length)), 
                        neighbour=unlist(nbs.list))

head(nbs.pairs)

##        zipcode neighbour
## 946011   94601     94501
## 946012   94601     94602
## 946013   94601     94605
## 946014   94601     94606
## 946015   94601     94621
## 946016   94601     94619    

Aultman

Thank you very much jbaums. This is what I was looking for. :-) However I could not find the shape files for KY, MS, NC, SC, TN and VA. Where should I be able to find them. Thank you very much. – Corey

I'm not sure that they are available for all states individually at geocommons, but you could try this file, which I imagine has zip codes for all states. (I haven't checked, since it's a 500MB download that I don't particularly need.) – Aultman

Also note the difference between ZIP codes and ZIP code tabulation areas. The latter are supplied in the above shapefile. – Aultman

Yes I do want the pairs. This is something I may spend some time on. – Bendy

@Corey Also: within the former (ZIP Codes) there are Zip Areas (~33,000) and Post Office/Large Value Customers (~10,000). The former are shapes, and the latter are points. Important to know the difference with analysis like this. – Naga

You can download more shapefiles at the Census Bureau, but it looks like they'll only have the Zip Code Tabulation Areas. This post on GIS SE discusses finding zip codes and they had trouble finding (free) zip codes. – Calamint