data-raw/process_cdc.R

In schifferl/LagSelectionBias:

cdc_cols <- cols(
    Notes = col_skip(),
    `Age Group` = col_character(),
    `Age Group Code` = col_skip(),
    Gender = col_character(),
    `Gender Code` = col_skip(),
    Race = col_character(),
    `Race Code` = col_skip(),
    Year = col_integer(),
    `Year Code` = col_skip(),
    Deaths = col_integer(),
    Population = col_integer(),
    `Crude Rate` = col_skip()
)

cdc_1968 <- read_tsv("inst/extdata/Compressed Mortality, 1968-1978.txt",
                     col_types = cdc_cols, n_max = 924, na = "Not Applicable")
cdc_1979 <- read_tsv("inst/extdata/Compressed Mortality, 1979-1998.txt",
                     col_types = cdc_cols, n_max = 1680, na = "Not Applicable")
cdc_1999 <- read_tsv("inst/extdata/Compressed Mortality, 1999-2015.txt",
                     col_types = cdc_cols, n_max = 1888, na = "Not Applicable")

cdc_data <-
    full_join(cdc_1968, cdc_1979) %>%
    full_join(cdc_1999) %>%
    rename(Age_Group = `Age Group`) %>%
    rename(Sex = Gender) %>%
    mutate(Race = gsub("American Indian or Alaska Native", "Other Race",
                       Race)) %>%
    mutate(Race = gsub("Asian or Pacific Islander", "Other Race", Race)) %>%
    select(Year, Age_Group, Race, Sex, Deaths, Population)