RE Conf 2016 Requirements Engineering in Agile Analytics 02. März 2016 Dr. Claudia Schindler, Felix Löw
Agenda 1 2 3 4 5 Munich Re Setting the context: Challenges for RE in Analytics Capturing the business context: The Reporting Information System Providing the right data: Data Sets and Data Flows Experiences and lessons learnt
Unser integriertes Geschäftsmodell Starke Marken im Verbund Munich Re (Gruppe)* Rückversicherung Munich Health Erstversicherung Corporate Insurance Partner Great Lakes Reinsurance (UK) PLC KA Köln.Assekuranz Agentur GmbH MSF Pritchard Syndicate 318 Temple Insurance Company Watkins Syndicate Belgium Assetmanagement * Die Darstellung erhebt keinen Anspruch auf Vollständigkeit und gibt nicht die genauen Beteiligungsverhältnisse REConf 2016 wieder. - Requirements Engineering in Agile Analytics / 02.03.2016
Kompetenzfelder von Munich Re Was uns von anderen unterscheidet Maßgeschneiderte Lösungen und Effizienz Service und Knowhow Sicherheit und Verlässlichkeit Wir übernehmen veränderte und komplexe Risiken erweitern die Grenzen der Versicherbarkeit, entwickeln innovative Deckungskonzepte. Wir bieten erstklassige Modellierung und maßgeschneiderte Deckungen, effiziente Rückversicherung von Standardrisiken, hohe Kapazität pro Risiko-Exponierung. Wir bieten risikoorientierte Services wie Underwriting-Tools, z.b. Nathan und MIRA, Wissenstransfer in Kundenseminaren, effiziente Kooperation über connect.munichre.com. Wir unterstützen Produktentwicklungen. Wir ermöglichen attraktives Kapitalmanagement. Transfer von Risiken an den Kapitalmarkt, Risikomanagement für Kapitalanlagen (ALM), Capital Relief, Solvency II Consulting.
In allen Märkten präsent Amelia Atlanta Chicago Columbus Hartford Montreal Philadelphia New York Princeton San Francisco Toronto Vancouver Munich London Madrid Malta Milan Moscow Paris Zurich Beijing Calcutta Dubai Hong Kong Kuala Lumpur Mumbai Seoul Shanghai Singapore Taipei Tokyo Bogotá Buenos Aires Caracas Mexico Santiago de Chile São Paulo Accra Cape Town Johannesburg Nairobi Port Louis Auckland Melbourne Sydney
Setting the context: Challenges for RE in Analytics 2
Setting the context - It s all about data!? How is this achieved? How is this consistently documented? L T E System A System B System C
In RE we have to tackle several challenges Business Process Usage requirements and contextual environment Domain Models Data Sources Roles and goals What is the business context? Frequency / Criticality What data is required? Logics Key Figures Personas Device How is data consumed? Use Cases UX Report Dashboard Interface Iterative Frequent user feedback Incremental Business Value Risk Driven
Traceability structure in Business Intelligence Business Processes Traceability User Stories Use Cases Reports Domain Model Data Sets Slices Data Sets
Capturing the business context The Reporting Information System 3
A structured evaluation of all existing reporting content has been achieved What is the business context? Who is using which report for what purpose in which location? Reporting Information System Location Reports Business Owner Business Process Business Process Step Actor
We are able to analyse our reporting portfolio from different perspectives
A structured evaluation of all existing reporting content has been achieved Which report is containing which data? What data is required? Key Figures Characteristics Logics Query Information System
Providing Data - Data Flows and Data Sets 4
Existing specification is complex and confusing What data is required? Domain Model Domain Model No clear responsibility for Excel documents No documentation for business
New form of specification is data centric Domain Model Data flow Domain Model Data flow per data set Requirements Engineers responsible Readable by business NFR Spec + per Report
Data sets are the basic entities of the reporting domain model providing business value Data Set Data Set: A collection of data belonging together from a business perspective, e.g. a class in a domain model
Data must be transformed from the source model to the target model to provide data sets Source Report
Every extraction process can be detailed by a drill down Reused source system domain model class Logical reporting data set ReinsuranceCoverageMgmt::Coverage + broker: Broker [0..1] = NULL + brokerage: Percent = 0% + cessionlimit: Amount + comment: Comment + coveragetype: CoverageType + eventlimit: Amount + id: OverallUniqueID {readonly} + internalid: int = NULL + isunlimited: bool = TRUE + largelosslimit: Amount + leadingbroker: Broker [0..1] + limit: Amount [0..1] = NULL + name: BoundedString + ordinalnumberwithincontractperiod: int «reporting» + isspecialtreaty: bool = FALSE + protectedshare: Percent = NULL + renewalstatus: RenewalStatus = NotActive + underwritingstatus: UnderwritingStatus = NotActive Extract relevant fields from TPS Coverage :: TPSCov erage - attachmentpoint: Amount - broker: Broker - coverageid: OverallUniqueID - coveragetype: CoverageType - isspecial: bool - limit: Amount - name: String - protectedshare: Percent - renewalstatus: RenewalStatus - underwritingstatus: UnderwritingStatus ReinsuranceCov eragemgmt::nonproportionalreinsurancecov erage + annualaggregatedeductible: Amount + annualaggregatedeductibletype: AnnualAggregateDeductibleType + annualaggregatelimit: Amount + attachmentpoint: Amount
Every transformation process can be detailed by a drill down
But additional documents are needed How is data consumed? Parameter Screen Selection Report Initial Layout and drill down attributes Report Overview Document: Report Specification Links to all relevant documents and specification of other topics (e.g. report specific NFR, authorization, )
Experience and lessons learnt 5
Our Experiences Project Team Test and Developer Data flows per data set help to slice requirements for iterative development Data sets used to structure test cases Unambiguous specification by using data flows instead of word documents If no source domain model exists, you can also start from the technical model Easy to read and understand, drill down to required level of detail helps Business Feedback Think they will like to review this kind of documentation The Business Reporting Repository enables business to manage Report portfolio Our Feedback Quality check of source system domain models Shows other projects which data are available and their structure The Business Reporting Repository supports a business value driven agile planning
Thank you for listening Ken Collier: Agile Analytics, Addison Wesley 2012 Dr. Claudia Schindler cschindler@munichre.com Felix Löw floew@munichre.com Ralph Hughes: Agile Data Warehousing for the Enterprise: A Guide for Solution Architects and Project Leaders 2015