Generali - Fraud Analytics
Download as PPTX, PDF2 likes1,266 views
This document discusses Generali Group, an international insurance company, and its efforts to detect insurance fraud. It provides an overview of Generali Group, describing it as one of the leading insurers in the world with operations in over 60 countries. It then discusses the problem of insurance fraud and some common types. It outlines Generali's solution to fraud detection, which involves using a graph database and applying techniques like rule-based analysis, social network analysis, and machine learning to processed insurance claims data.
Generali - Fraud Analytics
- 1. © Generali Deutschland Schadenmanagement GmbH Fraud Analytics Berlin, 27th February 2018 Neo4j GraphTour Thorsten Gomann
- 2. © Generali Deutschland Schadenmanagement GmbH 1 Agenda Generali Group Insurance Fraud Fraud Detection Solution based on neo4j Fraud Analytics / 2018-02-27 / Special Claims Department
- 3. © Generali Deutschland Schadenmanagement GmbH 2 One of the leading insurers in the world 1 Fortune Global 500 International Generali Group at a glance one of the world‘s 50 largest companies1 more than 60 countries in the world 74,000 employees Pioneer in Smart Insurance Generali Group 530 billion Euro Assets under Management 70.5 billion Euro in premiums (2016) over 66% from outside Italy Fraud Analytics / 2018-02-27 / Special Claims Department
- 4. © Generali Deutschland Schadenmanagement GmbH 3 Presence of the international Generali Group Through Generali Employee Benefits, the Group operates in more than 100 countries Insurance operations of Generali Group Branches: New York, United Kingdom, Hong Kong, Tokyo, Dubai, Panama Other countries in which Europ Assistance operates Fraud Analytics / 2018-02-27 / Special Claims Department
- 5. © Generali Deutschland Schadenmanagement GmbH 4 Generali in Germany as part of the worldwide Group GENERALI IN GERMANY IS THE SECOND-LARGEST COUNTRY UNIT 23% of the premiums of the Generali Group € 16.2 bn premiums in insurance business € 847 m operating result (FY2016) 12,400 employees Branche: United Kingdom 23 % Fraud Analytics / 2018-02-27 / Special Claims Department
- 6. © Generali Deutschland Schadenmanagement GmbH 5 Agenda Generali Group Insurance Fraud Fraud Detection Solution based on neo4j Fraud Analytics / 2018-02-27 / Special Claims Department
- 7. © Generali Deutschland Schadenmanagement GmbH 6 Insurance Fraud Exploitation Fictitious Claims Staged Claims Fraud Analytics / 2018-02-27 / Special Claims Department
- 8. © Generali Deutschland Schadenmanagement GmbH 7 Agenda Generali Group Insurance Fraud Fraud Detection Solution based on neo4j Fraud Analytics / 2018-02-27 / Special Claims Department
- 9. © Generali Deutschland Schadenmanagement GmbH 8 Approaches for Data Driven Fraud Detection Rules based on experts knowledge leading to probability score Create new (temporary) scenarios Rulesystem Recognition of claims participants in definedscenarios Features derived from network Social Network Analysis File-Metadata Manipulation Data File Forensics Textrecognition and interpretation • Keywords • Connections Semantic Analysis Linear Regression Logistic Regression Decision Tree Random Forest Gradient Boosting Neural Network Support Vector Machines Supervised Learning Anomalyrecognition Peer-Group Analysis Clustering Unsupervised Learning Fraud Analytics / 2018-02-27 / Special Claims Department
- 10. © Generali Deutschland Schadenmanagement GmbH 9 Agenda Generali Group Insurance Fraud Fraud Detection Solution based on neo4j Fraud Analytics / 2018-02-27 / Special Claims Department
- 11. © Generali Deutschland Schadenmanagement GmbH Generic Process for Fraud Detection based on Analytics 10 Fraud Analytics / 2018-02-27 / Special Claims Department
- 12. © Generali Deutschland Schadenmanagement GmbH Technical Integration 11 Operative Sources Stage Analytics View Graph-Model Fraud Analytics Fraud Analytics / 2018-02-27 / Special Claims Department {RESTful API}
- 13. © Generali Deutschland Schadenmanagement GmbH Administration for automated Scenarios 12 Fraud Analytics / 2018-02-27 / Special Claims Department
- 14. © Generali Deutschland Schadenmanagement GmbH Casetool Frontend 13 Fraud Analytics / 2018-02-27 / Special Claims Department
- 15. © Generali Deutschland Schadenmanagement GmbH Fraud Analytics / 2018-02-27 / Special Claims Department Thorsten Gomann Data Scientist Fraud / Recovery / Litigation Generali Deutschland Schadenmanagement GmbH Gereonswall 68 50670 Köln phone: 0221/1636-5645 thorsten.gomann@generali.com 14
Editor's Notes
- #10: Hinweis im Dezember auf AIXAM Datenbestand geprüft Regel erstellt. Erster Treffer in 1. Woche nach Regeleinführung. Unsupervised Learning Fahrertypen -> Höheres Alter mit hochwertigen Fahrzeugen aber geringeres Einkommen => vorsichtige Fahrer === evtl. Kasko Alle Altersklassen billige Fahrzeuge mit geringem Einkommen => hohe Schadenhäufigkeit === wittert Chance mit Rostlaube noch etwas Geld zu machen
- #13: Durchstich leichter Aufbau Performance leichte Erweiterbarkeit