setwd("C:/challenge_data")
data_set <- read.csv("Data_set.csv", header=TRUE)
table(data_set$MATE_OCCP_NAME_G)
occp_name_to_code <- function(row) {
  if(row=="1차산업 종사자") 
    row = 1
  else if(row=="2차산업 종사자")
    row = 2
  else if(row=="3차산업 종사자")
    row = 3
  else if(row=="*")
    row = 4
  else if(row=="고소득 전문직")
    row = 5
  else if(row=="공무원")
    row = 6
  else if(row=="기업/단체 임원")
    row = 7
  else if(row=="기타")
    row = 8
  else if(row=="단순 노무직")
    row = 9
  else if(row=="단순 사무직")
    row = 10
  else if(row=="사무직")
    row = 11
  else if(row=="예체능계 종사자")
    row = 12
  else if(row=="운전직")
    row = 13
  else if(row=="자영업")
    row = 14
  else if(row=="전문직")
    row = 15
  else if(row=="주부")
    row = 16
  else if(row=="학생")
    row = 17
  else
    row = 18
}
data_set$MATE_OCCP_NAME_G <- sapply(data_set$MATE_OCCP_NAME_G, occp_name_to_code)
data_set$MATE_OCCP_NAME_G <- unlist(data_set$MATE_OCCP_NAME_G)
table(data_set$MATE_OCCP_NAME_G)

data_set$OCCP_NAME_G <- sapply(data_set$OCCP_NAME_G, occp_name_to_code)
data_set$OCCP_NAME_G <- unlist(data_set$OCCP_NAME_G)
table(data_set$OCCP_NAME_G)

table(data_set$LT1Y_PEOD_RATE)
lt1y_peod_rate_to_code <- function(row) {
  if(row=="10미만") 
    row = 10
  else if(row=="20미만")
    row = 20
  else if(row=="30미만")
    row = 30
  else if(row=="40미만")
    row = 40
  else if(row=="50미만")
    row = 50
  else if(row=="60미만")
    row = 60
  else if(row=="90미만")
    row = 90
  else if(row=="90이상")
    row = 100
  else 
    row = 0
}
data_set$LT1Y_PEOD_RATE <- sapply(data_set$LT1Y_PEOD_RATE, lt1y_peod_rate_to_code)
data_set$LT1Y_PEOD_RATE <- unlist(data_set$LT1Y_PEOD_RATE)
table(data_set$LT1Y_PEOD_RATE)

#SKT Memgership grade to code
table(data_set$TEL_MBSP_GRAD)
tel_mgsp_grad_to_code <- function(row) {
  if(row=="E")
    row = 1
  else if(row=="Q")
    row = 2
  else if(row=="R")
    row = 3
  else if(row=="W")
    row = 4
  else 
    row = 0
}
data_set$TEL_MBSP_GRAD <- sapply(data_set$TEL_MBSP_GRAD, tel_mgsp_grad_to_code)
data_set$TEL_MBSP_GRAD <- unlist(data_set$TEL_MBSP_GRAD)
table(data_set$TEL_MBSP_GRAD)

table(data_set$CBPT_MBSP_YN)
yn_to_10 <- function(row) {
  if(row=="Y")
    row = 1
  else if(row=="N")
    row = 0
  else 
    row = 2
}
data_set$CBPT_MBSP_YN <- sapply(data_set$CBPT_MBSP_YN, yn_to_10)
data_set$CBPT_MBSP_YN <- unlist(data_set$CBPT_MBSP_YN)
table(data_set$CBPT_MBSP_YN)