Duplicate Payments
In addition to the rule-based duplicate payment detection dab AnalyticSuite offers also the possibility to enhance the output using machine learning by flagging the output of the analytic and hence reducing the number of false positives in a continous approach. The audit step is named AP_DuplicatePayments_Enhanced in dab Nexus Content Studio.
Flagging the results
To make use of the of the machine learning algorithm, the results need to flagged manually. To realise that two tables are required:
- AP_DuplicatePaymentsFlagged
This table is generated every run and contains the entries which need to be flagged and also the results of the machine learning algorithm whether this is estimated to be a false-positive. The following columns are relevant here:Column
Description
Target_Variable
This field should be the manual assessment of the user, if this is a duplicate payment or not, the following values are accepted:
- No Duplicate Payment
- True Duplicate Payment
AI_Assessment
Result of the AI algorithm whether this is a duplicate payment
AI_Potential
Probability that this is a true duplicate payment in the range from 0.0 to 1.0
- AP_DuplicatePaymentsPool
This table contains the previously flagged duplicate payments saved for the usage in the next analytic run. It will be continuely updated at every run with the newly flagged entries. If you delete the table you will loose all your progress on already flagged tables.
For further information and details regarding the columns and their reference to the underlying data refer to the analytic step documentation provided by your contact at dab.
Setting up the analytic in dab Nexus for continous execution
The AI-enhanced duplicate payment analytic is meant for continous execution not as one-time run. The following steps should be considered if running it in a continous scenario. Refer to the dab Nexus documentation for details.
- Create a data strategy that always uses the same database for every run
- Create a analytic group containing AP_DuplicatePayments_Enhanced
- Setup an analytic task using the created data strategy and analytic group
- You are good to go!
Avoiding execution failures
For proper operation the Analytic AP_DuplicatePayments needs to have the following columns in Content Studio selected, if you don't selected them you will get a crash when running this analytic:
- c_UpdateKey
- c_UniqueId
- BSAK_AUGDT
- BSAK_WAERS
- T001_WAERS
- BSAK_BUKRS
- BSAK_GJAHR
- c_Posting_Status
- c_BKPF_USNAM_DEPARTMENT
- c_Clearing_Department
- BSAK_ZTERM
- LFA1_SORTL
- LFA1_LAND1
- c_Status_PO
- LFA1_KTOKK
- SKAT_TXT20
- BSAK_BELNR
- BSAK_SGTXT
- BSAK_HKONT
- c_FinalDueDate
- LFA1_NAME1
- LFA1_ORT01
- BSAK_XBLNR
- BSAK_BUDAT
- BSAK_CPUDT
- BKPF_USNAM
- c_BKPF_USNAM_Clearing
- BKPF_TCODE
- c_BKPF_TCODE_Clearing
- BKPF_AWTYP
- BSAK_BLART
- c_BKPF_BLART_Clearing
- BKPF_DBBLG
- BSAK_ZFBDT
- BKPF_STBLG
- BSAK_MWSKZ
- BKPF_CPUTM
- c_BSAK_DMBTR_LC
- c_BSAK_DMBTR_RC
- BSAK_BLDAT
- BSAK_ZLSCH
