Insurance Data Science conference 2023
The conference took place at Bayes Business School, City, University of London, 15 - 16 June 2023
15 June 2023
09:15 - 10:15 Room B200: Keynote 1 (Chair: Andreas Tsanakas)
- Luca Baldassarre (Swiss Re): Responsible AI trade-offs in Insurance
10:15 - 11:15 Regular Session 1
Room B200: Machine learning and predictive modelling (Chair: Ron Richman)
- Munir Hiabu (University of Copenhagen): On functional decompositions, post-hoc machine learning explanations and fairness
- Mario V. Wüthrich (RiskLab, ETH Zurich): Isotonic recalibration under a low signal-to-noise ratio
- Can Baysal (Munich Re): Two-step Bayesian hyperparameter optimisation to efficiently build insurance market price models
Room B103: Advances in mortality modelling (Chair: Pietro Millossovich)
- Salvatore Scognamiglio (University of Naples): Accurate and Explainable Mortality Forecasting with the LocalGLMnet
- Mike Ludkovski (University of California Santa Barbara): Expressive Mortality Models through Gaussian Process Compositional Kernels
- Asmik Nalmpatian (Department of Statistics - LMU Munich, Allianz SE): Modern Machine Learning approaches in mortality modeling considering the impact of COVID-19
11:15 - 11:40 Coffee break
11:40 - 12:40 Lightning Session 1
Room B200: Stream 1 (Chair: Jürg Schelldorfer)
- Aurelien Couloumy (Novaa-Tech): Neural generative techniques for synthetic data creation in insurance: context and use case
- Thao Nguyen & Davide de March (Markel International): Sentence similarity models to develop a new risk appetite tool
- Navarun Jain (Lux Actuaries & Consultants): Saving the World: Predictive Early Warning Systems for Conflict Risk using Neural Networks
- Yafei (Patricia) Wang (Lloyd’s of London): Machine Learning and XAI for underwriting
- Nuzhat Jabinh (FRSA Ethics in AI consultant): Data is not neutral: ethics and AI
- *NN (Dupro Advisory) & Paul King (University of Leicster): AI Risk: How much should we care?
Room B103: Stream 2 (Chair: Diego Zappa)
- Mark Shoun (Ledger Investing): Fitting Development and Tail Models Jointly via Mixture Modeling
- Sebastian Calcetero-Vanegas (University of Toronto): A Credibility Index Approach for Effective a Posteriori Ratemaking with Large Insurance Portfolios
- Chris Halliwell & Cynon Sonkkila (Markel): Log-Normal-Pareto: A Case Study
- Gabriele Pittarello (University of Rome): Chain Ladder Plus: a versatile approach for claims reserving
- Mick Cooney (Describe Data): A Bayesian Approach to Customer Lifetime Value
- Tim Edwards (Howden Tiger): Estimating return periods for extreme events - a frequenstist and Bayesian perspective
12:40 - 13:40 Lunch
13:40 - 14:40 Regular Session 2
Room B200: NLP case studies (Chair: Davide de March)
- Jürg Schelldorfer (Swiss Re): Actuarial Applications of Natural Language Processing Using Transformers: Case Studies for Using Text Features in an Actuarial Context
- Paola Gasparini (Bupa): Advanced analytics and machine learning to identify fraudulent health insurance claims
- Bavo D.C. Campo (KU Leuven): Insurance fraud network data simulation machine: Generating synthetic fraud network data sets to develop and to evaluate insurance fraud detection strategies
Room B103: Fairness & explainability (Chair: Andreas Tsanakas)
- Olivier Côté (Université Laval): Causal Inference and Fairness in Insurance Pricing
- James Ng (Trinity College Dublin): Generalized Bayesian Inference with Fairness Constraints
- Deniz Günaydin-Bulut (Swiss Re): Why this claim? Incorporating local model explainability in a reinsurance setting
14:40 - 15:40 Panel discussion (Room B200)
Algorithmic underwriting: the future of specialty insurance? (Chair: John Ng, RGA)
- Davide Burlon (Principal, Insurance Solutions. Munich Re)
- Mick Cooney (CTO, Describe Data)
- Dana Cullen (Senior Associate, SCOR Ventures)
- Melanie Zhang (Head of Algorithmic Pricing at Ki)
15:40 - 16:00 Coffee break
16:00 - 17:00 Regular Session 3
Room B200: Challenges in modelling insurance data (Chair: Mick Cooney)
- Sindre Henriksen (Eika Forsikring): Implementing ML Ops in insurance: a case study using a complex, multi-model Customer Lifetime Value system
- Zhiyu Quan (University of Illinois at Urbana-Champaign): Imbalanced learning for insurance using modified loss functions in tree-based models
- Freek Holvoet (KU Leuven): Neural networks for insurance pricing with frequency and severity data: a benchmark study from data preprocessing steps to technical tariff
Room B103: Telematics & graphs (Chair: Mario Wüthrich)
- Xenxo Vidal-Llana (Universitat de Barcelona): Non-crossing neural network quantile regression estimation for driving data with telematics
- Marco De Virgilis (Arch Insurance): Territorial Ratemaking and Graph Theory
- Diego Zappa (Università Cattolica del Sacro Cuore): Estimating the road accident risk of a road network
17:00 - 18:00 Keynote 2 (Room B200) (Chair: Ioannis Kyriakou)
- Mark Sellors (Data Orchard): APIs and the future of data science
19:00 Conference dinner
- Ironmongers’ Hall, Shaftesbury Place, Barbican, London EC2Y 8AA
16 June 2023
09:15 - 10:15 Room B200: Keynote 3 (Chair: Mathias Lindholm)
- Rosalba Radice (Bayes Business School, City, University of London): A unifying and flexible bivariate copula regression framework
10:15 - 11:15 Lightning Session 2
Room B200: Stream 1 (Chair: Grainne McGuire)
- Patrick Hogan (PartnerRe, Zürich): ...whatever remains, however improbable, must be a bug
- Roland Schmid (Mirai Solutions): Py-shiny for Reinsurance: ready or not, here we come
- Priyank Shah (Lane Clark & Peacock LLP): Embedding data science in reserving
- William Mesquita (TROVADORES D´EQUAÇÕES LDA): Fully automated ETL Process Using Azure
- Bence Zaupper (Finalyse): Optimisation and automation of capital projections in insurance
- Amin Karbassi (Axa): Application of Information Retrieval (IR) for automation of Risk Engineering research within unstructured data
Room B103: Stream 2 (Chair: Mike Ludkovski)
- Despoina Makariou (University of St. Gallen): A bivariate mixed Poisson claim count regression model with varying dispersion and shape
- Lina Palmborg (Stockholm University): Reinforcement learning in search of optimal premium rules
- Bernard Wong (University of New South Wales): Machine Learning with High-Cardinality Categorical Features in Actuarial Applications
- Guillaume Biessy (LinkPact & Sorbonne Université): Revisiting Whittaker-Henderson Smoothing
- Agni Orfanoudaki (Saïd Business School, University of Oxford): Algorithmic Insurance
- Sukrita Singh (Saïd Business School, University of Oxford): Algorithmic Insurance: A Conformal Prediction Framework
11:15 - 11:45 Coffe break
11:45 - 12:45 Regular Session 4
Room B200: Risk modelling (Chair: Markus Gesmann)
- Claudio Giorgio Giancaterino (Intesa SanPaolo Vita): Earthquakes Risk Modelling with Quantile Approach
- Nicholas Robert (DeNexus): Vine Copulas for Systemic Cyber Risk Modelling
- Jacky Poon (nib Travel Insurance): From Chain Ladder to Probabilistic Neural Networks for Claims Reserving
Room B103: Tree based models in insurance (Chair: Munir Hiabu)
- Henning Zakrisson (Stockholm University): Multi-Parametric Gradient Boosting Machines with Non-Life Insurance Applications
- Mathias Lindholm (Stockholm University): Local bias adjustment, duration-weighted probabilities, and automatic construction of tariff cells
- Arthur Maillart (Detralytics): Distill knowledge of additive tree models into GAMs
12:45 - 13:45 Lunch
13:45 - 14:45 Lightning Session 3
Room B200: Stream 1 (Chair: Salvatore Scognamiglio)
- Karol Maciejewski (Milliman): The fast and the fabulous. Harnessing GPU power for high-performance life insurance computations
- Markus Gesmann (Insurance Capital Markets Research): Measuring daily value creation of global specialty (re)insurance
- Guillaume Attard (Ageoce Solutions): Evolution of the Soil Wetness Index (SWI) in France: Analysis with Google Earth Engine
- Shirley Ng (Vantage Risk): A Practitioner Guide to Marginal Pricing - Pricing with Portfolio Impact in Mind
- George Wright (Vounder Analytics): Is the Lloyd’s insurance market ready for an open-source capital modelling framework?
- Annette Hofmann (University of Cincinnati): Modeling Underwriting Cycles in Property-Casualty Insurance: The Impact of Catastrophic Events
Room B103: Stream 2 (Chair: Rui Zhu)
- Philipp Ratz (Université du Québec à Montréal): Solving censored regression problems using a multitask approach
Matteo Crisafulli (Università di Roma, La Sapienza): A neural network approach for selecting efficient reinsurance strategies- Michelle Dong (The Australian National University): Forecasting Mortality by cause with Zero Death Counts
- Valeria D’Amato (University of Salerno): Explore latent factors of longevity trends with frailty-based stochastic models
- Marie Michaelides (Université du Québec à Montréal): Individual claims reserving with dependent censored data
14:45 - 14:55 Room B200: Closing comments (Markus Gesmann)
Gold Sponsor
Jumping Rivers combines data science consultancy and knowledge transfer with provision of managed software to support businesses in gaining invaluable insights from their data.
Silver Sponsors
Posit: The open source data science company.
Mirai Solutions: Smarter analytics - better decisions
Markel: Bold ideas. Honest actions.
Carl H. Lindner III Center for Insurance and Risk Management