Check in update U-SQL script for scoring
This commit is contained in:
yiwsun
Родитель
a24b004399
Коммит
dbdf849e3a
|
|
@ -0,0 +1,173 @@
|
|||
REFERENCE ASSEMBLY [ExtR];
|
||||
|
||||
// R script to run
|
||||
DEPLOY RESOURCE @"/Samples/Scripts/FraudDetection-Scoring.R";
|
||||
|
||||
// Pretrained model file
|
||||
DEPLOY RESOURCE @"/Samples/Output/training/trainedmodel-rxbtree.rds";
|
||||
|
||||
// Input data for scoring (without headers)
|
||||
DECLARE @TransactionsInputFile string = @"/Samples/Data/transactions.csv";
|
||||
DECLARE @AccountsInputFile string = @"/Samples/Data/accounts.csv";
|
||||
|
||||
// Reference Tables
|
||||
DECLARE @AggTableFile string = @"/Samples/Output/training/aggtable.csv";
|
||||
DECLARE @risktable_ipstateInputFile string = @"/Samples/Output/training/riskTable_ipState.csv";
|
||||
DECLARE @risktable_ipPostalCodeInputFile string = @"/Samples/Output/training/riskTable_ipPostalCode.csv";
|
||||
DECLARE @risktable_ipCountryInputFile string = @"/Samples/Output/training/riskTable_ipCountry.csv";
|
||||
DECLARE @risktable_transactionIPaddressInputFile string = @"/Samples/Output/training/riskTable_transactionIPaddress.csv";
|
||||
DECLARE @risktable_paymentBillingPostalCodeInputFile string = @"/Samples/Output/training/riskTable_paymentBillingPostalCode.csv";
|
||||
DECLARE @risktable_shippingPostalCodeInputFile string = @"/Samples/Output/training/riskTable_shippingPostalCode.csv";
|
||||
DECLARE @risktable_accountPostalCodeInputFile string = @"/Samples/Output/training/riskTable_accountPostalCode.csv";
|
||||
|
||||
// Output files
|
||||
DECLARE @OutputFile string = @"/Samples/Output/prediction.csv";
|
||||
DECLARE @PartitionCount int = 100;
|
||||
|
||||
@Transactions = EXTRACT accountID string, transactionID string, transactionAmountUSD double, transactionCurrencyCode string, localHour int?,
|
||||
transactionScenario string, transactionType string, transactionIPaddress double, ipState string, ipPostalCode string,
|
||||
ipCountry string, isProxyIPNullable bool?, browserLanguage string, paymentInstrumentType string, cardType string,
|
||||
paymentBillingPostalCode string, paymentBillingState string, paymentBillingCountryCode string, shippingPostalCode string, shippingState string,
|
||||
shippingCountry string, cvvVerifyResult string, digitalItemCount int, physicalItemCount int, transactionDateTime string
|
||||
FROM @TransactionsInputFile
|
||||
USING Extractors.Csv(silent: true);
|
||||
|
||||
@Transactions = SELECT accountID, transactionID, transactionAmountUSD, transactionCurrencyCode, (int)(localHour ?? 0) AS localHour,
|
||||
transactionScenario, transactionType, transactionIPaddress, ipState, ipPostalCode,
|
||||
ipCountry, (bool) (isProxyIPNullable ?? false) AS isProxyIP, browserLanguage, paymentInstrumentType, cardType,
|
||||
paymentBillingPostalCode, paymentBillingState, paymentBillingCountryCode, shippingPostalCode, shippingState,
|
||||
shippingCountry, cvvVerifyResult, digitalItemCount, physicalItemCount, transactionDateTime FROM @Transactions;
|
||||
|
||||
@Accounts =
|
||||
EXTRACT accountID string, accountPostalCode string, accountState string, accountCountry string, accountAge string,
|
||||
isUserRegistered string, paymentInstrumentAgeInAccount string
|
||||
FROM @AccountsInputFile
|
||||
USING Extractors.Csv(silent: true);
|
||||
|
||||
// Join Transactions with Accounts
|
||||
@Trans_Account =
|
||||
SELECT @Transactions.*, accountPostalCode, accountState, accountCountry, accountAge, isUserRegistered, paymentInstrumentAgeInAccount
|
||||
FROM @Transactions
|
||||
LEFT JOIN @Accounts
|
||||
ON @Transactions.accountID == @Accounts.accountID;
|
||||
|
||||
|
||||
// Join Reference table of last 1day/30day aggregation features (account-level)
|
||||
@AggTable =
|
||||
EXTRACT accountID string, sumPurchaseAmount1Day double, sumPurchaseCount1Day int, sumPurchaseAmount30Day double, sumPurchaseCount30Day int
|
||||
FROM @AggTableFile
|
||||
USING Extractors.Csv(silent: true);
|
||||
|
||||
@Trans_Account_Agg =
|
||||
SELECT @Trans_Account. *,
|
||||
sumPurchaseAmount1Day ?? 0 AS sumPurchaseAmount1Day,
|
||||
sumPurchaseCount1Day ?? 0 AS sumPurchaseCount1Day,
|
||||
sumPurchaseAmount30Day ?? 0 AS sumPurchaseAmount30Day,
|
||||
sumPurchaseCount30Day ?? 0 AS sumPurchaseCount30Day
|
||||
FROM @Trans_Account
|
||||
LEFT JOIN
|
||||
@AggTable
|
||||
ON @Trans_Account.accountID == @AggTable.accountID;
|
||||
|
||||
|
||||
// Load Risk tables
|
||||
@Risktable_ipCountry = EXTRACT ipCountry string, ipCountry_risk double
|
||||
FROM @risktable_ipCountryInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_ipState = EXTRACT ipState string, ipState_risk double
|
||||
FROM @risktable_ipstateInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_ipPostalCode = EXTRACT ipPostalCode string, ipPostalCode_risk double
|
||||
FROM @risktable_ipPostalCodeInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_transactionIPaddress = EXTRACT transactionIPaddress double, transactionIPaddress_risk double
|
||||
FROM @risktable_transactionIPaddressInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_paymentBillingPostalCode = EXTRACT paymentBillingPostalCode string, paymentBillingPostalCode_risk double
|
||||
FROM @risktable_paymentBillingPostalCodeInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_shippingPostalCode = EXTRACT shippingPostalCode string, shippingPostalCode_risk double
|
||||
FROM @risktable_shippingPostalCodeInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
@Risktable_accountPostalCode = EXTRACT accountPostalCode string, accountPostalCode_risk double
|
||||
FROM @risktable_accountPostalCodeInputFile
|
||||
USING Extractors.Csv();
|
||||
|
||||
|
||||
// Assign risks
|
||||
@Trans_Account_Agg_Risk =
|
||||
SELECT @Trans_Account_Agg. *,
|
||||
ipCountry_risk ?? 0 AS ipCountry_risk,
|
||||
ipState_risk ?? 0 AS ipState_risk,
|
||||
ipPostalCode_risk ?? 0 AS ipPostalCode_risk,
|
||||
transactionIPaddress_risk ?? 0 AS transactionIPaddress_risk,
|
||||
paymentBillingPostalCode_risk ?? 0 AS paymentBillingPostalCode_risk,
|
||||
shippingPostalCode_risk ?? 0 AS shippingPostalCode_risk,
|
||||
accountPostalCode_risk ?? 0 AS accountPostalCode_risk
|
||||
FROM @Trans_Account_Agg
|
||||
LEFT JOIN
|
||||
@Risktable_ipCountry
|
||||
ON @Trans_Account_Agg.ipCountry == @Risktable_ipCountry.ipCountry
|
||||
LEFT JOIN
|
||||
@Risktable_ipState
|
||||
ON @Trans_Account_Agg.ipState == @Risktable_ipState.ipState
|
||||
LEFT JOIN
|
||||
@Risktable_ipPostalCode
|
||||
ON @Trans_Account_Agg.ipPostalCode == @Risktable_ipPostalCode.ipPostalCode
|
||||
LEFT JOIN
|
||||
@Risktable_transactionIPaddress
|
||||
ON @Trans_Account_Agg.transactionIPaddress == @Risktable_transactionIPaddress.transactionIPaddress
|
||||
LEFT JOIN
|
||||
@Risktable_paymentBillingPostalCode
|
||||
ON @Trans_Account_Agg.paymentBillingPostalCode == @Risktable_paymentBillingPostalCode.paymentBillingPostalCode
|
||||
LEFT JOIN
|
||||
@Risktable_shippingPostalCode
|
||||
ON @Trans_Account_Agg.shippingPostalCode == @Risktable_shippingPostalCode.shippingPostalCode
|
||||
LEFT JOIN
|
||||
@Risktable_accountPostalCode
|
||||
ON @Trans_Account_Agg.accountPostalCode == @Risktable_accountPostalCode.accountPostalCode;
|
||||
|
||||
// Add boolean features, drop redundant columns
|
||||
@Trans_Account_Agg_Risk_Binary = SELECT accountID, transactionID, transactionAmountUSD, transactionCurrencyCode, localHour, transactionScenario, transactionType,
|
||||
isProxyIP, browserLanguage, paymentInstrumentType, cardType,
|
||||
paymentBillingState, paymentBillingCountryCode, shippingState,
|
||||
shippingCountry, cvvVerifyResult, digitalItemCount, physicalItemCount, transactionDateTime,
|
||||
accountState, accountCountry, accountAge, isUserRegistered, paymentInstrumentAgeInAccount, sumPurchaseAmount1Day,
|
||||
sumPurchaseCount1Day, sumPurchaseAmount30Day, sumPurchaseCount30Day, ipState_risk, ipPostalCode_risk, ipCountry_risk,
|
||||
transactionIPaddress_risk, paymentBillingPostalCode_risk, shippingPostalCode_risk, accountPostalCode_risk,
|
||||
(transactionAmountUSD >150 ? true : false) AS is_highAmount,
|
||||
(paymentBillingPostalCode==accountPostalCode ? false : true) AS acct_billing_postalCode_mismatchFlag,
|
||||
(paymentBillingCountryCode==accountCountry ? false : true) AS acct_billing_country_mismatchFlag,
|
||||
(shippingPostalCode==accountPostalCode ? false : true) AS acct_shipping_postalCode_mismatchFlag,
|
||||
(shippingCountry==accountCountry ? false : true) AS acct_shipping_country_mismatchFlag,
|
||||
(shippingPostalCode==paymentBillingPostalCode ? false : true) AS shipping_billing_postalCode_mismatchFlag,
|
||||
(shippingCountry==paymentBillingCountryCode ? false : true) AS shipping_billing_country_mismatchFlag
|
||||
FROM @Trans_Account_Agg_Risk;
|
||||
|
||||
@ExtendedData =
|
||||
SELECT Extension.R.RandomNumberGenerator.GetRandomNumber(@PartitionCount) AS Par,
|
||||
*
|
||||
FROM @Trans_Account_Agg_Risk_Binary;
|
||||
|
||||
|
||||
|
||||
@RScriptOutput =
|
||||
REDUCE @ExtendedData
|
||||
ON Par
|
||||
PRODUCE Par,
|
||||
accountID string,
|
||||
transactionID string,
|
||||
transactionAmountUSD double,
|
||||
score double
|
||||
READONLY Par
|
||||
USING new Extension.R.Reducer(scriptFile : "FraudDetection-Scoring.R", rReturnType : "dataframe");
|
||||
|
||||
|
||||
OUTPUT @RScriptOutput TO @OutputFile USING Outputters.Csv();
|
||||
|
||||
Загрузка…
Ссылка в новой задаче