R/collapse_01.R

In kamclean/collaborator: Scalable multi-centre research using R and REDCap

# collapse01--------------------------------

# Documentation
#' Collapse a group of binary variables (e.g. those generated by checkbox variables in REDCap) into summary columns.
#' @description Collapse a group of binary variables (e.g. those generated by checkbox variables in REDCap) into summary columns.
#' @param df Dataframe.
#' @param cols List of columnn names of binary variables desired to be collapsed.
#' @param prefix String to add to the start of all summary columns names.
#' @param suffix String to add to the end of all summary columns names.
#' @param remove Logical value to remove columns supplied to the "cols" parameter
#' @param output List of desired outputs: yesno (any values selected), n (number of values selected) and list (list of column names of all values selected).
#' @param binary0 List of all values corresponding to "0" (No) in the binary variable (default = 0 or no)
#' @param binary1 List of all values corresponding to "1" (Yes) in the binary variable (default = 1 or yes).
#' @return Dataframe with up to 3 additional columns: yesno (any values selected), n (number of values selected) and list (list of column names of all values selected).
#' @import dplyr
#' @import tibble
#' @import tidyr
#' @importFrom tidyselect all_of
#' @export

# Function
collapse01 <- function(df, cols, prefix=NULL, suffix = NULL,
                       output = c("list", "yesno", "n"), remove = TRUE,
                       binary0 = c(0, "0", "No", "no"), binary1 = c(1, "1", "Yes", "yes")){

  require(tibble);require(dplyr);require(tidyr);require(tidyselect)
  #clean df
  out <- df %>%
    dplyr::select_at(tidyselect::all_of(cols)) %>%
    dplyr::mutate_all(function(x){ifelse(x %in% binary0, NA, x)}) %>%
    tibble::rowid_to_column() %>%

    # wranging to get colname inserted (inefficent)
    tidyr::pivot_longer(cols = -rowid) %>%
    dplyr::mutate(value = ifelse(value %in% binary1, name, value)) %>%
    tidyr::pivot_wider(id_cols = rowid) %>%

    # create columns of interest (yesno, n, and list)
    tidyr::unite(col = "list", -rowid, na.rm = T, sep = ", ") %>%
    dplyr::mutate(list = ifelse(list=="", NA, list)) %>%
    dplyr::mutate(n = stringr::str_count(list, ", ")+1) %>%
    dplyr::mutate(n = ifelse(is.na(n)==T, 0, n)) %>%
    dplyr::mutate(yesno = factor(ifelse(n==0, "No", "Yes"), levels = c("No", "Yes"))) %>%
    dplyr::select(all_of(output)) %>%

    dplyr::rename_all(function(x){paste0(if(is.null(prefix)==T){""}else{prefix},
                                         x,
                                         if(is.null(suffix)==T){""}else{suffix})})

  out <- tibble::as_tibble(dplyr::bind_cols(df, out))

  if(remove==TRUE){out <- out %>% dplyr::select(-any_of(cols))}

  return(out)}