CODEWITHSUNDEEP
rlapplypurrrtidycensus
Mox
Mox

Reputation: 541

Use purrr() to rather than lapply for arbitrary city/county pairs in tidycensus?

I've got a giant lapply running to download a file of data. But it seems clumsy. But mapply does not seem right, as I don't want all state/county combinations. I hear good things about map(). Can anyone furnish an example of how I might use purrr() command 'map' for the following code? 

library(tidycensus)
library(sf)
mykey<-"youhavetogetyourownimafraid"
#variables to test out the function#############
x<-"06"
y<-"073"
z<-"2000"
setwd("N:/Dropbox/_BonesFirst/149_Transit_Metros_BG_StateSplit_by_R")
##################now the actual function#########################
get_Census <- function(x,y,z) {
  name<-paste0("transitmetro_",x,"_",y,"_",z)
  name
  namefile<-get_decennial(geography = "block group",
                variables = "P001001",
                sumfile = "sf1",
                key = mykey,
                state = x, county = y,year = z,
                geometry = TRUE)
  st_write(namefile, paste0(name,".shp")) #tidycensus version of write OGR
}  

#now, for all of them
CO<-c("013")
tibble04_10<-lapply(CO,get_Census,x="04",z="2000")
CO<-c("067","073","113")
tibble06_10<-lapply(CO,get_Census,x="06",z="2000")
CO<-c("005","031","035")
tibble08_10<-lapply(CO,get_Census,x="08",z="2000")
CO<-c("037","053","123")
tibble27_10<-lapply(CO,get_Census,x="27",z="2000")
CO<-c("119")
tibble37_10<-lapply(CO,get_Census,x="37",z="2000")
CO<-c("085","113","121","201")
tibble48_10<-lapply(CO,get_Census,x="48",z="2000")
CO<-c("035") #SLCO, utah
tibble49_10<-lapply(CO,get_Census,x="49",z="2000")
CO<-c("033","053") #King co, Seattle
tibble53_10<-lapply(CO,get_Census,x="53",z="2000")

EDIT

get_Census <- function(x,y,z) {
  name<-paste0("transitmetro_",x,"_",y,"_",z)
  name
  namefile<-get_decennial(geography = "block group",
                          variables = "P001001",
                          sumfile = "sf1",
                          key = mykey,
                          state = x, county = y,year = z,
                          geometry = TRUE)
  st_write(namefile, paste0(name,".shp")) #tidycensus version of write OGR
}  

CO_list <- list(c("013"),
                c("067","073","113"),
                c("005","031","035"),
                c("037","053","123"),
                c("119"),
                c("085","113","121","201"),
                c("035"),
                c("033","053"))

x_list <- c("04", "06", "08", "27", "37", "48", "49", "53")
z_list <- c("2000", "2000", "2000", "2000", "2000", "2000", "2000", "2000")
# BUILD LIST OF OBJECTS
tibble_list <- Map(function(CO, x, z) lapply(CO, function(i) get_Census(i, x, z)), 
                   CO_list, x_list, z_list)
# NAME LIST OF OBJECTS: tibble04_10, tibble06_10, tibble08_10, ...
tibble_list <- setNames(tibble_list, paste0("tibble", x_list, "_10"))
print(tibble_list)

Yields:

Getting data from the 2000 decennial Census Error : Result 1 is not a length 1 atomic vector In addition: Warning messages: 1: '004' is not a valid FIPS code for counties in Georgia
2: '004' is not a valid FIPS code for counties in Georgia
Show Traceback
Rerun with Debug Error in gather_(data, key_col = compat_as_lazy(enquo(key)), value_col = compat_as_lazy(enquo(value)), : unused argument (-NAME)

Upvotes: 2

Views: 278

Answers (2)

Parfait
Parfait

Reputation: 107747

Re-consider base R's apply family, namely Map (wrapper to mapply) and lapply which I hear good things about. Simply build equal length lists to pass into a nested function call.

CO_list <- list(c("013"),
                c("067","073","113"),
                c("005","031","035"),
                c("037","053","123"),
                c("119"),
                c("085","113","121","201"),
                c("035"),
                c("033","053"))

x_list <- c("04", "06", "08", "27", "37", "48", "49", "53")
z_list <- c("2000", "2000", "2000", "2000", "2000", "2000", "2000", "2000")

# BUILD LIST OF OBJECTS
tibble_list <- Map(function(CO, x, z) lapply(CO, function(i) get_Census(i, x, z)), 
                   CO_list, x_list, z_list)

# NAME LIST OF OBJECTS: tibble04_10, tibble06_10, tibble08_10, ...
tibble_list <- setNames(tibble_list, paste0("tibble", x_list, "_10"))

Also, since z_list is redundant being all the same, you can shorten:

tibble_list <- Map(function(CO, x) lapply(CO, function(i) get_Census(i, x, z=2000)), 
                   CO_list, x_list)

Upvotes: 3

camille
camille

Reputation: 16871

You can use map from purrr to map over a list of your county characteristics. One way to do this is a list of lists, such as:

fips_yrs <- list(
    sl_co = list(x = "49", y = "035", z = 2000),
    king_co = list(x = "53", y = "053", z = 2000)
)

Then map will map over each county, and you can pull out its info by name with [[1]].

map(fips_yrs, ~get_Census(x = .$x[[1]], y = .$y[[1]], z = .$z[[1]]))

FYI, if all you need is the shapefiles, tidycensus uses functions from tigris to download its shapefiles, so you could just call a tigris function directly.

Upvotes: 2

Related Questions