BEGIN:VCALENDAR VERSION:2.0 METHOD:PUBLISH CALSCALE:GREGORIAN PRODID:-//Unimelb add to calendar// v1.0//EN BEGIN:VEVENT UID:5572615@unimelb.edu.au DTSTAMP:20260616T100000 DTSTART:20260616T100000 DTEND:20260616T110000 SUMMARY:Xindi (Cindy) Fang PhD Completion Seminar DESCRIPTION:Presenter: Xindi FangSupervisors: Xueyuan Wu and David PittTitle: The Study of Modern Technologies and Big Data Analytical Methods in InsuranceDay/Time: Tuesday 16 June 2026, 10 - 11am Australian Eastern Standard Time (AEST)Venue: FBE-315Zoom Link: https://unimelb.zoom.us/j/81702984811?pwd=gF1isscauM50aslEAZ4F4bSxNqypTb.1(Password: 115419)Abstract:The InsurTech industry has developed rapidly over the past decade, but its long-term growth now faces challenges from global uncertainty, regulatory change, and concerns about sustainable profitability. This thesis first examines the evolution of InsurTech through a data-driven review of both academic research and industry practice. Based on more than 20,000 academic and practitioner publications collected from 2000 to March 2025, the study applies a keyword extraction framework to compare publication growth, emerging topics, business-line development, and gaps between academic and practitioner perspectives.Building on this review, the thesis focuses on usage-based insurance, particularly telematics-based automobile insurance, as one of the earliest and most important InsurTech applications. Telematics insurance uses driving behaviour data collected through connected devices to support more individualised actuarial valuation. Despite its potential, its implementation remains limited by practical challenges. A key difficulty is the information delay problem: reliable behaviour-based assessment usually requires several months of driving records, creating a cold-start issue for new policyholders and insurers with limited historical telematics data. To address this issue, the thesis considers self-assessment driving questionnaires as an accessible early-stage proxy before stable telematics-based risk evaluation becomes available. These measures are studied alongside traditional demographic and driving experience variables, as well as real-life telematics summaries, in short-horizon dynamic prediction settings.The thesis then develops a dynamic daily risk scoring framework for telematics insurance. This framework supports the transition towards a “manage how you drive” system by combining past risk experience with evolving daily telematics-based driving behaviour through a hybrid MINAR+XGB framework. The model addresses sparsity in risk frequencies and extends the analysis to multivariate outcomes across different severity levels, allowing the framework to capture temporal dependence, nonlinear telematics effects, and dependence across risk categories.Overall, the thesis links macro-level evidence on InsurTech development with detailed actuarial applications in telematics-based usage-based insurance, supporting more adaptive and data-informed insurance valuation under practical real-world constraints. X-ALT-DESC;FMTTYPE=text/html:

Presenter: Xindi Fang

Supervisors: Xueyuan Wu and David Pitt

Title: The Study of Modern Technologies and Big Data Analytical Methods in Insurance

Day/Time: Tuesday 16 June 2026, 10 - 11am Australian Eastern Standard Time (AEST)

Venue: FBE-315

Zoom Link: https://unimelb.zoom.us/j/81702984811?pwd=gF1isscauM50aslEAZ4F4bSxNqypTb.1
(Password: 115419)

Abstract:

The InsurTech industry has developed rapidly over the past decade, but its long-term growth now faces challenges from global uncertainty, regulatory change, and concerns about sustainable profitability. This thesis first examines the evolution of InsurTech through a data-driven review of both academic research and industry practice. Based on more than 20,000 academic and practitioner publications collected from 2000 to March 2025, the study applies a keyword extraction framework to compare publication growth, emerging topics, business-line development, and gaps between academic and practitioner perspectives.

Building on this review, the thesis focuses on usage-based insurance, particularly telematics-based automobile insurance, as one of the earliest and most important InsurTech applications. Telematics insurance uses driving behaviour data collected through connected devices to support more individualised actuarial valuation. Despite its potential, its implementation remains limited by practical challenges. A key difficulty is the information delay problem: reliable behaviour-based assessment usually requires several months of driving records, creating a cold-start issue for new policyholders and insurers with limited historical telematics data. To address this issue, the thesis considers self-assessment driving questionnaires as an accessible early-stage proxy before stable telematics-based risk evaluation becomes available. These measures are studied alongside traditional demographic and driving experience variables, as well as real-life telematics summaries, in short-horizon dynamic prediction settings.

The thesis then develops a dynamic daily risk scoring framework for telematics insurance. This framework supports the transition towards a “manage how you drive” system by combining past risk experience with evolving daily telematics-based driving behaviour through a hybrid MINAR+XGB framework. The model addresses sparsity in risk frequencies and extends the analysis to multivariate outcomes across different severity levels, allowing the framework to capture temporal dependence, nonlinear telematics effects, and dependence across risk categories.

Overall, the thesis links macro-level evidence on InsurTech development with detailed actuarial applications in telematics-based usage-based insurance, supporting more adaptive and data-informed insurance valuation under practical real-world constraints.

LOCATION: URL:https://fbe.unimelb.edu.au/economics/events/seminars/xindi-cindy-fang-phd-completion-seminar TRANSP:TRANSPARENT END:VEVENT END:VCALENDAR