荷兰合作银行利用合规的 Lakehouse 实现信用分析转型.pdf

1、Transforming Credit AnalyticsWith a Compliant Lakehouse at RabobankTaras Chaikovskyi Floris Hendriks10 June 20252Taras ChaikovskyiSr.Resident Solutions Architect DatabricksFloris HendriksProduct ManagerRabobankDisclaimerDisclaimer:This presentation should not be reported as representing the views of

2、 the Rabobank or its employees.The views expressed are those of the authors and do not necessarily reflect those of the Rabobank.Agenda4Business context-Credit Analytics at RaboFlorisWhy UC?FlorisTransition from Hive to Unity CatalogTarasUnity catalog upgrade frameworkTarasWrap-up,next steps and Q&A

3、Floris&TarasRabobankSource:Rabobank-at-a-glance-2024.pdf125+year old49 000+FTE2024Source:Rabobank-global-presence-2024.pdfCredit AnalyticsCredit modelling predicts default risk and potential losses Multiple types of models:Internal ratings-based(IRB),IRFS9,From data collection,model creation,deploym

4、ent and monitoringBetter credit decisions,risk management,and regulatory complianceImportant credit risk parameters used in key operations of the bankUsed at scaleStrongly regulated,it should be transparent and in-controlData&AIAcceptanceMonitoringReportingPricingCredit data lakehouseData must be ti

5、mely,complete,accurate,and adaptable to evolving needs Long horizon of historical dataIntegration needed of diverse data sources Run at In-control at scale Supported by business processesnot just technical solutions8GoalsSimple model life cycleData CollectionModel TrainingModel Validation/ApprovalMo

6、del UseModel MonitoringData challengeBridging different systemsABCDEODFODFTimeTimeDefinitionDefinitionSystemSystemDoD XDoD YDoD ZSystem ASystem BSystem CAEActual history in reportingBack casted regulatory compliant definitionDQ issue(needs to be solved for consistent back casted output)Different sys