Advertisement
Guest User

Untitled

a guest
Dec 16th, 2017
102
0
Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
text 5.80 KB | None | 0 0
  1. ################# Data Description ############
  2. #account = (4500 objects in the file ACCOUNT.ASC) ? each record describes static
  3. #characteristics of an account,
  4. #permanent order = (6471 objects in the file ORDER.ASC) ? each record describes
  5. #characteristics of a payment order,
  6. #transaction = (1056320 objects in the file TRANS.ASC) ? each record describes one
  7. #transaction on an account,
  8. #loan = (682 objects in the file LOAN.ASC) ? each record describes a loan granted for a
  9. #given account,
  10. ###############################################
  11.  
  12. GetFrequencyValuesEn <- function()
  13. {
  14. return (list(MONTHLY = "Monthly", WEEKLY = "Weekly", TRANSACTION = "AfterTransaction"))
  15. }
  16.  
  17.  
  18. GetFrequencyValuesCz <- function()
  19. {
  20. return (list(MONTHLY = "POPLATEK MESICNE", WEEKLY = "POPLATEK TYDNE", TRANSACTION = "POPLATEK PO OBRATU"))
  21. }
  22.  
  23.  
  24. ReadAccountsDataFrame <- function()
  25. {
  26. accounts.df <- (ReadDataFrameFromFilepath("account.asc"))
  27.  
  28. # Normalizes date values
  29. accounts.df$date <- sapply(accounts.df$date, GetBirthDateYyMmDd)
  30.  
  31. # Normalizes statement frequency values
  32. fv.cz <- GetFrequencyValuesCz()
  33. fv.en <- GetFrequencyValuesEn()
  34.  
  35. accounts.df$frequency <- mapvalues(accounts.df$frequency,
  36. from = c(fv.cz$MONTHLY, fv.cz$WEEKLY, fv.cz$TRANSACTION),
  37. to = c(fv.en$MONTHLY, fv.en$WEEKLY, fv.en$TRANSACTION))
  38.  
  39. return (as.data.frame(accounts.df))
  40. }
  41.  
  42. GetTypeValuesCz <- function()
  43. {
  44. return (list(CREDIT = "PRIJEM", DEBIT = "VYDAJ", CHOICE = "VYBER"))
  45. }
  46.  
  47. GetTypeValuesEn <- function()
  48. {
  49. # Debit is called withdrawal on the pdf but debit is a more common name
  50. return (list(CREDIT = "Credit", DEBIT = "Debit", CHOICE = "Choice"))
  51. }
  52.  
  53. GetOperationValuesCz <- function()
  54. {
  55. return (list(CARDDEBIT = "VYBER KARTOU", CASHCREDIT = "VKLAD", CASHDEBIT = "VYBER",
  56. OTHERBANKCREDIT = "PREVOD Z UCTU", OTHERBANKDEBIT = "PREVOD NA UCET"))
  57. }
  58.  
  59. GetOperationValuesEn <- function()
  60. {
  61. # remittance to another bank = sum of money sent to other institution
  62. # collection = credit from another institution
  63. return (list(CARDDEBIT = "CardDebit", CASHCREDIT = "CashCredit", CASHDEBIT = "CashDebit",
  64. OTHERBANKCREDIT = "OtherBankCredit", OTHERBANKDEBIT = "OtherBankDebit"))
  65. }
  66.  
  67. ReadTransactionsDataFrame <- function()
  68. {
  69. trans.df <- ReadDataFrameFromFilepath("trans.asc")
  70.  
  71. # Normalizes operation values
  72. op.cz <- GetOperationValuesCz()
  73. op.en <- GetOperationValuesEn()
  74.  
  75. trans.df$operation <- mapvalues(trans.df$operation,
  76. from = c(op.cz$CARDDEBIT, op.cz$CASHCREDIT, op.cz$CASHDEBIT, op.cz$OTHERBANKCREDIT, op.cz$OTHERBANKDEBIT),
  77. to = c(op.en$CARDDEBIT, op.en$CASHCREDIT, op.en$CASHDEBIT, op.en$OTHERBANKCREDIT, op.en$OTHERBANKDEBIT))
  78.  
  79. # Normalizes date values
  80. trans.df$date <- sapply(trans.df$date, GetBirthDateYyMmDd)
  81.  
  82. tp.cz <- GetTypeValuesCz()
  83. tp.en <- GetTypeValuesEn()
  84.  
  85. trans.df$type <- mapvalues(trans.df$type,
  86. from = c(tp.cz$CREDIT, tp.cz$DEBIT, tp.cz$CHOICE),
  87. to = c(tp.en$CREDIT, tp.en$DEBIT, tp.en$CHOICE))
  88.  
  89. ks.cz <- GetKsymbolValuesCz()
  90. ks.en <- GetKsymbolValuesEn()
  91.  
  92. trans.df$k_symbol <- mapvalues(trans.df$k_symbol,
  93. from = c(ks.cz$INSURANCE, ks.cz$HOUSEHOLD, ks.cz$LOAN, ks.cz$PENSION, ks.cz$STATEMENT, ks.cz$SANCTION, ks.cz$INTEREST),
  94. to = c(ks.en$INSURANCE, ks.en$HOUSEHOLD, ks.en$LOAN, ks.en$PENSION, ks.en$STATEMENT, ks.en$SANCTION, ks.en$INTEREST))
  95.  
  96. return (as.data.frame(trans.df))
  97. }
  98.  
  99. GetLoanStatusEn <- function()
  100. {
  101. return (list(FINISHEDOK = "ContractFinishedOk", FINISHEDDEFAULT = "ContractFinishedDefault", RUNNINGOK = "RunningOk", RUNNINGDEBT = "RunningInDebt"))
  102. }
  103.  
  104. GetLoanStatusLetters <- function()
  105. {
  106. return (list(FINISHEDOK = "A", FINISHEDDEFAULT = "B", RUNNINGOK = "C", RUNNINGDEBT = "D"))
  107. }
  108.  
  109. ReadLoansDataFrame <- function()
  110. {
  111. loans.df <- (ReadDataFrameFromFilepath("loan.asc"))
  112.  
  113. # Normalizes date values
  114. loans.df$date <- sapply(loans.df$date, GetBirthDateYyMmDd)
  115.  
  116. ls.en <- GetLoanStatusEn()
  117. ls.lt <- GetLoanStatusLetters()
  118.  
  119. loans.df$status <- mapvalues(loans.df$status,
  120. from = c(ls.lt$FINISHEDOK, ls.lt$FINISHEDDEFAULT, ls.lt$RUNNINGOK, ls.lt$RUNNINGDEBT),
  121. to = c(ls.en$FINISHEDOK, ls.en$FINISHEDDEFAULT, ls.en$RUNNINGOK, ls.en$RUNNINGDEBT))
  122.  
  123. return (as.data.frame(loans.df))
  124. }
  125.  
  126. # k_symbol is a column in permanent order and also in transactions
  127. GetKsymbolValuesEn <- function()
  128. {
  129. return (list(INSURANCE = "Insurance", HOUSEHOLD = "Household", LEASING = "Leasing", LOAN = "Loan", PENSION = "Pension", STATEMENT = "StatementForPayment", SANCTION = "SanctionNegativeBalance", INTEREST = "InterestCredit"))
  130. }
  131.  
  132. GetKsymbolValuesCz <- function()
  133. {
  134. return (list(INSURANCE = "POJISTNE", HOUSEHOLD = "SIPO", LEASING = "LEASING", LOAN = "UVER", PENSION = "DUCHOD", STATEMENT = "SLUZBY", SANCTION = "SANKC. UROK", INTEREST = "UROK"))
  135. }
  136.  
  137. ReadPermanentOrdersDataFrame <- function()
  138. {
  139. orders.df <- (ReadDataFrameFromFilepath("order.asc"))
  140.  
  141. ks.cz <- GetKsymbolValuesCz()
  142. ks.en <- GetKsymbolValuesEn()
  143.  
  144. orders.df$k_symbol <- mapvalues(orders.df$k_symbol,
  145. from = c(ks.cz$INSURANCE, ks.cz$HOUSEHOLD, ks.cz$LEASING, ks.cz$LOAN),
  146. to = c(ks.en$INSURANCE, ks.en$HOUSEHOLD, ks.en$LEASING, ks.en$LOAN))
  147.  
  148. return (as.data.frame(orders.df))
  149. }
  150.  
  151. ReadCreditCardsDataFrame <- function()
  152. {
  153. cards.df <- (ReadDataFrameFromFilepath("card.asc"))
  154.  
  155. # Normalizes date values
  156. cards.df$issued <- sapply(cards.df$issued, GetBirthDateYyMmDd)
  157.  
  158. return (as.data.frame(cards.df))
  159. }
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement