R script to download, extract, and create the single year age, SEER U.S. County Population Data dataset

seer_single_year.r

library(tidyverse)
library(data.table)

# import, unzip, and read
# this is for IL, change the url to whichever version you want/need from here: 
# https://seer.cancer.gov/popdata/download.html
il_singleage = fread(
  "curl https://seer.cancer.gov/popdata/yr1969_2022.singleages/il.1969_2022.singleages.txt.gz | funzip", 
  colClasses = 'character',
  header = FALSE
) |>
  mutate(
    yr = str_sub(V1, 1, 4),
    state_abbv = str_sub(V1, 5, 6),
    state_fips = str_sub(V1, 7, 8),
    county_fips = str_sub(V1, 9, 11),
    registry = str_sub(V1, 12, 13),
    race = str_sub(V1, 14, 14),
    origin = str_sub(V1, 15, 15),
    sex = str_sub(V1, 16, 16),
    age = str_sub(V1, 17, 18),
    pop = as.numeric(str_sub(V1, 19, 27)),
    age_group = case_when(
      age == "00" ~ "0 years",
      age == "01" ~ "1 years",
      age == "02" ~ "2 years",
      age == "03" ~ "3 years",
      age == "04" ~ "4 years",
      age == "05" ~ "5 years",
      age == "06" ~ "6 years",
      age == "07" ~ "7 years",
      age == "08" ~ "8 years",
      age == "09" ~ "9 years",
      age == "10" ~ "10 years",
      age == "11" ~ "11 years",
      age == "12" ~ "12 years",
      age == "13" ~ "13 years",
      age == "14" ~ "14 years",
      age == "15" ~ "15 years",
      age == "16" ~ "16 years",
      age == "17" ~ "17 years",
      age == "18" ~ "18 years",
      age == "19" ~ "19 years",
      age == "20" ~ "20 years",
      age == "21" ~ "21 years",
      age == "22" ~ "22 years",
      age == "23" ~ "23 years",
      age == "24" ~ "24 years",
      age == "25" ~ "25 years",
      age == "26" ~ "26 years",
      age == "27" ~ "27 years",
      age == "28" ~ "28 years",
      age == "29" ~ "29 years",
      age == "30" ~ "30 years",
      age == "31" ~ "31 years",
      age == "32" ~ "32 years",
      age == "33" ~ "33 years",
      age == "34" ~ "34 years",
      age == "35" ~ "35 years",
      age == "36" ~ "36 years",
      age == "37" ~ "37 years",
      age == "38" ~ "38 years",
      age == "39" ~ "39 years",
      age == "40" ~ "40 years",
      age == "41" ~ "41 years",
      age == "42" ~ "42 years",
      age == "43" ~ "43 years",
      age == "44" ~ "44 years",
      age == "45" ~ "45 years",
      age == "46" ~ "46 years",
      age == "47" ~ "47 years",
      age == "48" ~ "48 years",
      age == "49" ~ "49 years",
      age == "50" ~ "50 years",
      age == "51" ~ "51 years",
      age == "52" ~ "52 years",
      age == "53" ~ "53 years",
      age == "54" ~ "54 years",
      age == "55" ~ "55 years",
      age == "56" ~ "56 years",
      age == "57" ~ "57 years",
      age == "58" ~ "58 years",
      age == "59" ~ "59 years",
      age == "60" ~ "60 years",
      age == "61" ~ "61 years",
      age == "62" ~ "62 years",
      age == "63" ~ "63 years",
      age == "64" ~ "64 years",
      age == "65" ~ "65 years",
      age == "66" ~ "66 years",
      age == "67" ~ "67 years",
      age == "68" ~ "68 years",
      age == "69" ~ "69 years",
      age == "70" ~ "70 years",
      age == "71" ~ "71 years",
      age == "72" ~ "72 years",
      age == "73" ~ "73 years",
      age == "74" ~ "74 years",
      age == "75" ~ "75 years",
      age == "76" ~ "76 years",
      age == "77" ~ "77 years",
      age == "78" ~ "78 years",
      age == "79" ~ "79 years",
      age == "80" ~ "80 years",
      age == "81" ~ "81 years",
      age == "82" ~ "82 years",
      age == "83" ~ "83 years",
      age == "84" ~ "84 years",
      age == "85" ~ "85+ years"
    )
  ) |>
  unite(fips, state_fips:county_fips, sep = "", remove = FALSE) |>
  select(-V1)