PI-1: Improving Agency Analytics
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Seth Goodchild
Panelists:
Justin Milam, Tom Heise, Seth Goodchild
M-9: How to Find Data-driven Insights When You Have No Data (or Just Want More)
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Rebecca Williams
Panelists:
Jessica Leong, Blake Berman
Keywords:
Data Quality (204),Data Mining (813),Data Management and Information (200),Ratemaking (101)
M-8: Inside the Black Box: Methods for ML Interpretation
Model transparency is essential to actuarial modeling, enabling the build of optimal models and the interpretation of model results for business use. The session will center around explanatory tools that communicate the predictive and interpretative power of analytic models. We will discuss common tools, such as SHAP and LIME, and how they are calculated and can be used to understand models that are typically "black box".
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Gaetan Veilleux
Panelists:
Megan Camanocha, Erool Palipane, Gaetan Veilleux
M-7: Historical versus Model Data
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Steven Phillips
Panelists:
Daniel Ward, Michael Tine
M-6: Bonus-Malus Scales Models for Predictive Modeling
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Brian Stoll
Panelists:
Jean-Philippe Boucher
M-5: Interpretability and Transparent Machine Learning
This session will discuss interpretation techniques for machine learning and AI. The session will focus on making machine learning models and their decisions interpretable. We will explore and compare different interpretation techniques (Feature Importance, Accumulated Local Effects, Partial Dependency Plots, Shapley Values, etc.). How do these techniques work? What are the advantages and disadvantages for each technique? How can their outputs be interpreted? Additionally, we will be discussing and comparing Shapley values between GBM vs Random Forest vs Layered GBM.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Don Hendriks
Panelists:
Liam McGrath, Rachael McNaughton
M-4: Mixture-of-Experts Models for Claim Frequency and Severity
Generalized linear models have been widely used for regression modelling in property and casualty insurance, but they may perform quite poorly when real data exhibit behaviours that violate their underlying assumptions, such as excessive zeros, multimodality and heavy tails. In this presentation, we introduce a new open-source statistical software tailor-made for actuarial applications which allows actuarial practitioners to model and analyze insurance loss frequency and severity using a nonlinear multivariate regression model. The model is very flexible, and it can fit any type of positive dataset, capture the dependency structure implied by the data, and is statistically implementable. Several distinctive features are included which cannot be achieved using existing software packages, including a wider coverage on frequency and severity distributions and their zero-inflated versions, parameter estimation under data censoring and truncation, and a collection of insurance ratemaking and reserving functions. The packages also provide several model evaluation and visualization functions to help users easily analyze the performance of the fitted model and interpret the model in insurance contexts. The presentation also contains a demonstration of how to use our software to analyze a real insurance dataset.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Mary Jo Kannon
Panelists:
Ian Weng Chan
Keywords:
Financial and Statistical Methods (700),Personal Lines (8888),Ratemaking (101),Reserving (130),Statistical Models and Methods (810)
M-3: Advanced Machine Learning in Real Life
Prior to a rate change implementation, it is common for pricing actuaries, product managers and other insurance leaders to compare various rate scenarios and choose the one that best aligns with business objectives and regulatory considerations. However, to fully understand the impact of a rate change on a book of business, customer demand should be considered. This session will discuss how advances in machine learning are being made to better solve these types of insurance problems. This will be a dynamic and interactive session where the audience will be able to provide guidance on model design and rate impact selection to see how all these pieces fit together.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Gary Wang
Panelists:
Michael Chen, Serhat Guven, Gary Wang
M-2: Penalized Regression Between Credibility and GBMs
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Steven Phillips
Panelists:
Guillaume Beraud-Sudreau, Mattia Casotto
M-1: The Practical Impact of Credibility Models
Penalized regressions such as Lasso and Elastic Net are becoming part of the standard toolkit for modelers and actuaries. The original adoption of penalized regression can be traced back to its appealing theoretical properties, such as the ability to blend GLM with credibility assumptions and better predictive power. When these methods are adopted for insurance modeling, they allow an actuarial modeler to intuitively see the relationships between penalization, credibility, parsimony, and stability. After a brief theoretical overview, this session will cover how and why these new techniques are being adopted by the industry from a modeler's practical perspective.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Concurrent Session
Moderators:
Steven Phillips
Panelists:
Thomas Holmes, James Riley, Jake Falandays, Steven Phillips
Keywords:
Statistical Models and Methods (810)
GS-2: IP (Inflationary Pressure)! IP (Industry Panel)! Hooray?
In the aftermath of COVID-19, the economy has been facing inflationary pressures that impact all types of ratemaking! Our panel of experts from across the Property & Casualty Industry will explore inflationary pressures in our industry, touching on Workers Compensation, Reinsurance, and Social Inflation!
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
General Session
Moderators:
Rebecca Williams
Panelists:
Stephanie Rabin, Jim Lynch, Sean Cooper, Rebecca Williams
Keywords:
Dynamic Risk Modeling (190),Extreme Event Modeling (932),Ratemaking (101),Reinsurance (610)
GS-1: All Hands on Deck: Risk Management and the US Navy
Whether you sit at a desk in an office on dry land or serve our country aboard a naval carrier far from shore, effective risk management is vital to success. Similar to the insurance industry, the US Navy aims to foster a proactive culture of excellence that encourages risk management, problem-solving and critical thinking. During this session, Admiral Bird will discuss how the US Navy uses effective risk management at all levels within the Department of the Navy.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Featured Speaker
Moderators:
Kathleen Ores Walsh
Panelists:
John Bird, Kathleen Ores Walsh
Keywords:
Enterprise Risk Management (150)
iCAS-5: Hands-on git/GitHub Tutorial
In this tutorial session, Brian will cover the basics of using version control with git and GitHub. You will understand the business case for introducing these tools into your actuarial workflow and how to get started. You will learn how git differs from other backup and change management procedures and the benefits this offers. You will then learn how to arrange a set of files, start making and tracking changes, and document the evolution of your work. Finally, you will push your work to the cloud and see how your contributions are visible and accessible to others.
Attendees should bring their laptops, with git installed and ready to use. Attendees will find it beneficial to watch the microlearning series on git and GitHub.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Workshop
Moderators:
Brian Fannin
Panelists:
Brian Fannin
iCAS-4: Latent Dirichlet Allocation (LDA) Topic Modeling in Python
Valuable data that could be used for business insights and decision-making is often only available in free form text fields. This presentation will review one method of converting text to more usable information by using Latent Dirichlet Allocation (LDA) topic modeling. This discussion will include an overview of where this method was applied to voicemail transcripts to quantify the most common reasons claims participants call claim representatives, with the results helping to prioritize IT investment in building out self-service functionality.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
iCAS Community of Practice Event
Panelists:
Denise Christophel
iCAS-3: Penalized Regression and Credibility Applications
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
iCAS Community of Practice Event
Panelists:
Thomas Holmes
iCAS-2: Data Governance: From Framework to Implementation Roundtable
In this Roundtable, we will review how data governance is defined in the context of different organizations and what challenges are faced by the Insurance industry. After reviewing a comprehensive framework, we will explore how insurance companies could implement it successfully.
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
iCAS Community of Practice Event
Panelists:
Yoram Gilboa
iCAS-1: The Logit-weighted Reduced Mixture of Experts (LRMoE) - A flexible modeling tool for actuarial applications
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
iCAS Community of Practice Event
Moderators:
Jamie Mills
Panelists:
Andrei Badescu, Ian Weng Chan
iCAS: iCAS Data Science & Analytics Forum
Source:
2023 Ratemaking, Product and Modeling Seminar
Type:
Workshop
Moderators:
Eric Krafcheck
Source:
2022 Annual Meeting
Type:
General Session
Moderators:
Steve Belden
Panelists:
Deniz Cultu ; Stephen Belden; Holly Price
Keywords:
Leadership
GS-3: Embedded Insurance and How It Could Change the Industry Landscape
As technology has evolved so have economies and market dynamics. Embedded insurance has entered the scene as an area with rapid growth potential in a shifting insurance distribution paradigm. This session will bring in thought leaders in the space to not only define embedded insurance but also explore the opportunities and challenges it could present for the industry.
Source:
2022 Annual Meeting
Type:
General Session
Moderators:
Adrian Jones
Panelists:
Joshua Snead; Michael Meeks; Cole Platt
Keywords:
Emerging Issues
GS-2: Key Risks and Opportunities: Today and Tomorrow
Source:
2022 Annual Meeting
Type:
General Session
Moderators:
Katey Walker
Panelists:
Sean Kevelighan; Dale Porfilio
Keywords:
Emerging Issues
GS-1: Chief Actuary Roundtable: A Discussion on Diversifying the Pipeline and Ensuring All Our Members Thrive
Source:
2022 Annual Meeting
Type:
General Session
Moderators:
Kathy Antonello
Panelists:
Bruce Gifford; Donna Glenn; David Cummings; Wesley Griffiths
Keywords:
Leadership
FS-1: Featured Speaker Address: Cathy O'Neil
CAS' Annual Meeting Featured speaker, Cathy O'Neil, PhD., will share her thoughts, analyses and concerns over the uses of data and algorithms to draw incomplete or inaccurate conclusions and what we as a profession can do to level the field. She will address algorithmic auditing in general, and about her framework called explainable fairness in particular.
*Introduced by Thomas Holmes, Akur8
** If you wish to participate via livestream/webcast, this requires a separate registration.
Source:
2022 Annual Meeting
Type:
Featured Speaker
Moderators:
Kathy Antonello
Panelists:
Cathy O'Neil
CS-9: Building the Actuary of the Future
The industry has evolved and so have the skills needed to be a successful actuary and risk expert. This session covers the hard and soft skills that an actuary should strive to possess to maximize their impact and value creation. We will also explore various growth opportunities that exist for actuaries and how those can be personally and professionally rewarding.
Source:
2022 Annual Meeting
Type:
Concurrent Session
Moderators:
Chelsea Adler
Panelists:
Dominic Lee; Rafael Costa; Josh Taub
Keywords:
Business Skills
CS-8: Autonomous Vehicles: Things an Actuary Should Know
In this session we will speak about autonomous vehicles, how they function, what data is collected and how we can think about quantifying the safety benefits of the technology.
Source:
2022 Annual Meeting
Type:
Concurrent Session
Moderators:
Tilia Gode
Panelists:
Tilia Gode; John Scanlon
Keywords:
Autonomous Vehicles