R/tidyPa.R

In europace-privatkredit/tidyPkuData: Tidy Europace loan data

#' product provider factor levels
#'
#' @description This function recodes factor levels in product provider vector
#' @param df The name of the dataframe you want to manipulate, agg defines whether regional product providers are aggregated or not
#' @return datatframe, tibble
#' @keywords product provider levels
#' @import dplyr, stringr
#' @export
#' @examples fun_tidyPa(df = df)

tidyPa <- function(df, aggregate = TRUE) {
  
  if (aggregate) {
    
    df <- df %>% 
      dplyr::mutate(produktanbieterid = dplyr::case_when(
        stringr::str_detect(produktanbieterid, pattern = 'PSD') ~ 'PSD',
        stringr::str_detect(produktanbieterid, pattern = 'SPK') | stringr::str_detect(produktanbieterid, pattern = 'KSK') ~ 'Sparkasse',
        stringr::str_detect(produktanbieterid, pattern = 'DSL') ~ 'DSL',
        stringr::str_detect(produktanbieterid, pattern = 'DKB') ~ 'DKB',
        stringr::str_detect(produktanbieterid, pattern = 'NETBANK') ~ 'Netbank',
        stringr::str_detect(produktanbieterid, pattern = 'TARGOBANK') ~ 'Targo',
        stringr::str_detect(produktanbieterid, pattern = 'SANTANDER_CONSUMER') ~ 'Santander-Consumer',
        stringr::str_detect(produktanbieterid, pattern = 'SKG') ~ 'SKG',
        stringr::str_detect(produktanbieterid, pattern = 'DIBA') ~ 'ING-Diba',
        stringr::str_detect(produktanbieterid, pattern = 'OAB') ~ 'OAB',
        stringr::str_detect(produktanbieterid, pattern = 'AUXMONEY') ~ 'AUXMONEY', 
        stringr::str_detect(produktanbieterid, pattern = 'SANTANDER_CARCREDIT') ~ 'Santander-Car',
        stringr::str_detect(produktanbieterid, pattern = 'COMMERZBANK') ~ 'Coba',
        stringr::str_detect(produktanbieterid, pattern = 'ALTE_LEIPZIGER') ~ 'Alte-Leipziger',
        stringr::str_detect(produktanbieterid, pattern = 'SUEDWESTKREDITBANK') ~ 'SWK',
        stringr::str_detect(produktanbieterid, pattern = 'VON_ESSEN_BANK') ~ 'VEB',
        TRUE~ 'nicht_vorhanden')) #%>% 
    # filter(produktanbieterid != 'Test')
    
    return(df)
    
  }
  
  else {
    
    df <- df %>% 
      dplyr::mutate(produktanbieterid = str_remove_all(produktanbieterid, pattern = '_RATENKREDIT')) %>% 
      dplyr::filter(produktanbieterid != 'ABRECHNUNGSBANK_RATENKREDIT')
    
    return(df)
  }
  
}