data-raw/caracteristique.R

In ngsanogo/corpoaccident: Base de données sur les accidents corporels de la circulation en France

## code to prepare `caracteristique` dataset goes here

library(conflicted)
library(data.table)
library(purrr)
library(dplyr)
library(lubridate)

caracteristique <- list.files("./data-raw/raw", pattern = "^caracteristiques", full.names = TRUE) %>%
  map(fread, colClasses = "character", encoding = "Latin-1", na.strings = c("", "NA")) %>%
  bind_rows() %>%
  as_tibble() %>%
  set_names(tolower) %>%
  mutate(
    an = case_when(
      nchar(an) == 4 ~ an,
      nchar(an) == 2 ~ paste0("20", an),
      nchar(an) == 1 ~ paste0("200", an),
      TRUE ~ NA_character_
    ),
    across(c(an, mois, jour), as.numeric),
    hrmn = case_when(
      nchar(hrmn) == 5 ~ hrmn,
      nchar(hrmn) == 4 ~ gsub(hrmn, pattern = "^(.{2})(.+)$", replacement = "\\1:\\2"),
      nchar(hrmn) == 3 ~ gsub(paste0("0", hrmn), pattern = "^(.{2})(.+)$", replacement = "\\1:\\2"),
      nchar(hrmn) == 2 ~ gsub(paste0("00", hrmn), pattern = "^(.{2})(.+)$", replacement = "\\1:\\2"),
      nchar(hrmn) == 1 ~ gsub(paste0("000", hrmn), pattern = "^(.{2})(.+)$", replacement = "\\1:\\2"),
      TRUE ~ NA_character_
    ),
    heure_accident = ymd_hm(paste0(an, "-", mois, "-", jour, " ", hrmn)),
    .after = hrmn,
    lum = case_when(
      lum %in% "1" ~ "Plein jour",
      lum %in% "2" ~ "Crépuscule ou aube",
      lum %in% "3" ~ "Nuit sans éclairage public",
      lum %in% "4" ~ "Nuit avec éclairage public non allumé",
      lum %in% "5" ~ "Nuit avec éclairage public allumé",
      TRUE ~ NA_character_
    ),
    agg = case_when(
      agg %in% "1" ~ "Hors agglomération",
      agg %in% "2" ~ "En agglomération",
      TRUE ~ NA_character_
    ),
    int = case_when(
      int %in% "1" ~ "Hors intersection",
      int %in% "2" ~ "Intersection en X",
      int %in% "3" ~ "Intersection en T",
      int %in% "4" ~ "Intersection en Y",
      int %in% "5" ~ "Intersection à plus de 4 branches",
      int %in% "6" ~ "Giratoire",
      int %in% "7" ~ "Place",
      int %in% "8" ~ "Passage à niveau",
      int %in% "9" ~ "Autre intersection",
      TRUE ~ NA_character_
    ),
    atm = case_when(
      atm %in% "1" ~ "Normale",
      atm %in% "2" ~ "Pluie légère",
      atm %in% "3" ~ "Pluie forte",
      atm %in% "4" ~ "Neige - grêle",
      atm %in% "5" ~ "Brouillard - fumée",
      atm %in% "6" ~ "Vent fort - tempête",
      atm %in% "7" ~ "Temps éblouissant",
      atm %in% "8" ~ "Temps couvert",
      atm %in% "9" ~ "Autre",
      TRUE ~ NA_character_
    ),
    col = case_when(
      col %in% "1" ~ "Deux véhicules - frontale",
      col %in% "2" ~ "Deux véhicules – par l’arrière",
      col %in% "3" ~ "Deux véhicules – par le coté",
      col %in% "4" ~ "Trois véhicules et plus – en chaîne",
      col %in% "5" ~ "Trois véhicules et plus - collisions multiples",
      col %in% "6" ~ "Autre collision",
      col %in% "7" ~ "Sans collision",
      TRUE ~ NA_character_
    ),
    gps = case_when(
      gps %in% "M" ~ "Métropole",
      gps %in% "A" ~ "Antilles (Martinique ou Guadeloupe)",
      gps %in% "G" ~ "Guyane",
      gps %in% "R" ~ "Réunion",
      gps %in% "Y" ~ "Mayotte",
      TRUE ~ NA_character_
    )
  )

usethis::use_data(caracteristique, overwrite = TRUE, compress = "bzip2", ascii = FALSE)