A study on approaches to developing customer risk profiles for underwriting using knowledge graph and multilabel classification

Nguyen, Khanh Van

A study on approaches to developing customer risk profiles for underwriting using knowledge graph and multilabel classification

Nguyen, Khanh Van

Permalink

Publication Type:: Thesis
Issue Date:: 2023

Open Access

Copyright Clearance Process

Recently Added
In Progress
Open Access

This item is open access.

Adobe PDF

Download thesisAdobe PDF (2.94 MB)

View statistics

Full metadata record

Field	Value	Language
dc.contributor.author	Nguyen, Khanh Van
dc.date.accessioned	2025-01-30T00:57:14Z
dc.date.available	2025-01-30T00:57:14Z
dc.date.issued	2023
dc.identifier.uri	http://hdl.handle.net/10453/184602
dc.description	University of Technology Sydney. Faculty of Engineering and Information Technology.	en_US.UTF-8
dc.description.abstract	As the insurance industry expands into new markets and mass customers, the volume of data required to process with regards to a customer's mortality risk has become significantly exponential. However, this process, based on the traditional underwriting process soon becomes less optimal and lacks personalisation for individual customers, especially when the process still heavily relies on manual intervention, leading to inconsistencies and biases. Hence, this research identifies challenges such as handling missing values, adopting a more data-driven approach, and ensuring model explainability. To tackle these issues, this study proposes the development of an Underwriting Knowledge Graph (UKG) combined with a multi-label classification model embedded that integrates all customer information received with the existing underwriting manual and explainable exclusion code system to formulate personalised customer risk profiles. The UKG utilises interconnected information pertaining to the customers, including historical data, and explainable exclusion to create customer risk profile, while offering valuable insights and potential correlations with specific exclusion codes. The study's deep dive into insurance domain-specific requirements, current research limitation within automated underwriting has led to the creation of the first-ever UKG, trained on real-life underwriting data thanks to the collaboration with our Australian industry partner. In addition to the UKG, the study also introduces a semi-automated novel method for maintaining the UKG that factors in the multi-label classification nature of the data outcome to provide explainable exclusion. The UKG as a data structure provides a comprehensive understanding of the insurance ecosystem, facilitates a representation of information on customer risk profiles, and enables explainable exclusion classifications.	en_US.UTF-8
dc.format	Thesis (MAnalytics)
dc.language.iso	en_US	en_US.UTF-8
dc.relation	https://opus.lib.uts.edu.au/bitstream/10453/184602/1/thesis.pdf
dc.rights	info:eu-repo/semantics/openAccess
dc.rights	The author owns the copyright in this thesis including all reproduction and reuse rights for the work. The work may not be altered without the permission of the copyright owner. Attribution is essential when quoting or paraphrasing from this thesis.
dc.rights	© 2023 Khanh Van Nguyen
dc.rights	au.edu.uts.lib/nph
dc.title	A study on approaches to developing customer risk profiles for underwriting using knowledge graph and multilabel classification	en_US.UTF-8
dc.type	Thesis
utslib.copyright.status	open_access	*

Abstract:

As the insurance industry expands into new markets and mass customers, the volume of data required to process with regards to a customer's mortality risk has become significantly exponential. However, this process, based on the traditional underwriting process soon becomes less optimal and lacks personalisation for individual customers, especially when the process still heavily relies on manual intervention, leading to inconsistencies and biases. Hence, this research identifies challenges such as handling missing values, adopting a more data-driven approach, and ensuring model explainability. To tackle these issues, this study proposes the development of an Underwriting Knowledge Graph (UKG) combined with a multi-label classification model embedded that integrates all customer information received with the existing underwriting manual and explainable exclusion code system to formulate personalised customer risk profiles. The UKG utilises interconnected information pertaining to the customers, including historical data, and explainable exclusion to create customer risk profile, while offering valuable insights and potential correlations with specific exclusion codes. The study's deep dive into insurance domain-specific requirements, current research limitation within automated underwriting has led to the creation of the first-ever UKG, trained on real-life underwriting data thanks to the collaboration with our Australian industry partner. In addition to the UKG, the study also introduces a semi-automated novel method for maintaining the UKG that factors in the multi-label classification nature of the data outcome to provide explainable exclusion. The UKG as a data structure provides a comprehensive understanding of the insurance ecosystem, facilitates a representation of information on customer risk profiles, and enables explainable exclusion classifications.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/184602