R/write_descriptives.R

In sarid-ins/saridr: An R package for common tasks related to market research analysis via R

#' Make descriptives of a list of numeric variables, and write it in excel
#' all descriptives in a single table
#'
#' @param x a tbl() that contains the variables of interest, keyed by ID variable
#' @param ... list of variables names that would appear in output
#' @param folder_name where would the output file will be saved (project's home by default).
#'
#' @return descriptives tbl
#'
#' @examples
#' data=tribble(~ID,~car_size,~height,
#'             "1",120,71,
#'              "2",110,81,
#'             "3",90,91,
#'              "4",80,120,
#'             "5",70,300,
#'              "6",60,1,
#'              "7",NA,80)
#' 
#' data %>%
#'    write_descriptives(car_size,height,folder_name = "R")

#'
#' @importFrom magrittr %>%
#' @importFrom dplyr enquos
#' @importFrom dplyr mutate
#' @importFrom dplyr rename
#' @importFrom dplyr count
#' @importFrom dplyr select
#' @importFrom dplyr everything
#' @importFrom dplyr group_by
#' @importFrom dplyr summarise
#' @importFrom openxlsx createWorkbook
#' @importFrom openxlsx addWorksheet
#' @importFrom openxlsx writeData
#' @importFrom openxlsx createStyle
#' @importFrom openxlsx saveWorkbook
#'
#' @export
#'


write_descriptives <- function(x,...,folder_name=getwd(),
                          name="descriptives_of_variables.xlsx"){
  #for all variable in variable list make a big table with n's
  
  vlist <- enquos(...)
  
  ntotal <- nrow(x)
  
  wb <- createWorkbook()
  hs1 <- createStyle(
    fgFill = "#DCE6F1", halign = "CENTER", textDecoration = "bold",
    border = "bottom"
  )
  addWorksheet(wb, "variables_descriptives")
  
  #name <- paste0("n_of_variables.csv")
  #cat(paste("printing-",name))
  
  reslist <- list()
  
  for (vname in vlist){
    reslist[[as.character(vname)[2]]] <- x %>%
        mutate(var_name=as.character(vname)[2]) %>%
        group_by(var_name) %>% 
        summarise(n=sum(!is.na(!!vname)),
                  missing=sum(is.na(!!vname)),
                  avg=round(mean(!!vname,na.rm=T),2),
                  median=round(median(!!vname,na.rm=T),2),
                  max=round(max(!!vname,na.rm=T),2),
                  min=round(min(!!vname,na.rm=T),2),
                  std=round(sd(!!vname,na.rm=T),2)) %>%
        # summarise(n=n()) %>% 
        # count(var,!!vname) %>%
        # rename(value=!!vname) %>% 
       
        mutate(percent=paste0(round((n/ntotal)*100),"%")) %>%  
        mutate(total=ntotal) 
  }
  
  x <- do.call(rbind, reslist) 
  
  writeData(wb, "variables_descriptives", x,
            colNames = F,
            rowNames = F,
            startCol = "A", startRow = 2,
            borders = "surrounding", borderColour = "black"
  )
  writeData(wb, "variables_descriptives", x,
            colNames = T,
            rowNames = F,
            startCol = "A", startRow = 1,
            borders = "surrounding", borderColour = "black",headerStyle = hs1
  )
  
  
  saveWorkbook(wb,file = paste0(folder_name,"/",name), overwrite = TRUE)
  
  
  return(x)
}