Speeches, Presentations and Workshops


CAS Reserve Variability Limited Attendance Seminar
November 7-9, 2018

With Mark Shapland, Miliman, Inc.
Philadelphia, PA

Insurance Data Concepts and Analytics Workshop and Reception
September 20, 2018
CAS and IDMA sponsoring
Jersey City, NJ

Intermediate R Workshop – CAS Loss Reserve Seiminar
September 5, 2018
This workshop covers the following topics:

  • Loops
  • User defined functions
  • Data pre-processing using tidyverse packages including:
  • tidyr
  • dplyr
  • ChainLadder library for reserving
  • Rmarkdown
  • Apps
  • A case study using open source data

International Congress of Actuaries
June 3, 2018
Berlin, Germany

Big Data Working Party II
The Big Data Working Party II is a follow-up to the work of the two previous working parties. It introduces frequently used predictive modeling methods. Arguably the most commonly used predictive modeling approach in insurance, is generalized linear models, or GLM. The working party uses GLM as a baseline against which to compare the other methods introduced. It is assumed readers are familiar with the method, due to its widespread use. While the working party papers do not provide an introduction to the theory and application of GLMs, readers are walked through the steps used to fit and assess a reasonable model (steps that are applicable to many other types of models) and provided R code for the steps. The working party focuses on introducing other approaches that largely originate from the statistics and machine learning disciplines. The working party was grouped into teams that worked on specific topics. The following techniques are covered:

Unsupervised Learning
Clustering K means Clustering Hierarchical Clustering Principal Components Analysis Principal Components Analysis
Supervised Learning
Some of the most popular non-GLM supervised learning methods are: Trees Ensemble Trees (randomForest) neural networks, and Multivariate Adaptive Regression Splines (MARS). In this round of the Big Data Working Party, Trees and MARS are introduced. The introduction assumes that the reader is familiar with ordinary least squares regression.

The CAS Institutes Predictive Modelers Practitioners Day
March 19, 2018
Data Preparation Issues
Presented recent developments in R tools that are helpful in data preparation for data scientists. R packages such as dplyr and caret are introduced, along with string processing procedures. Where possible examples use open source data.


CAS Ratemaking and Product Management Seminar
March 27-29, 2017
San Diego, CA
Text Mining


Casualty Actuarial Society Annual (CAS) Meeting – November 15-18, Philadelphia

  • Panelist in General Session on Actuaries versus Data Scientists in Predictive Modeling in Insurance
  • Moderator and speaker in session on Enterprise Risk Management research sponsored by the Joint Risk Management Section

Actuarial Research Conference
August 5-8, 2015, Toronto, Canada


PM-PA-6: Predictive Modeling for Actuaries Book Project

Predictive Modeling for Actuaries Book Project
Abstract: The editors embarked on a two-volume book project that incorporates a discussion of techniques in 20 Volume One chapters and case studies with data sets in 10-15 Volume Two chapters. Volume One was released in July 2014, and Volume Two is expected to be published in the fall of 2015.


2014 Casualty Actuarial Society Annual Meeting and Centennial Celebration

Participated in session on Predictive Modeling Applications in Actuarial Science, Volume One,
See 2013 presentations for downloads of material on the book.

2014 Casualty Loss Reserve Seminar, September 2014, San Diego

VR-1: Regression Based Testing of the Assumptions of Underlying Loss Development – See more at: http://www.casact.org/education/clrs/2014/index.cfm?fa=consess&Meeting=clrs14

The Committee on the Theory of Risk (COTOR) developed a syllabus and materials for a class on testing the assumptions underlying the estimates of loss development factors. Louise Francis was on the committee when the project was done. Many of the tests are regression based. This session provides a summary and key elements of the one day class including:

  • The data sets used
  • The Excel regression functions needed
  • The assumptions underlying chain ladder
  • How to test these assumptions
  • How to obtain and use the facilitator’s guide to the class

This session had a practical focus and the data sets are available. In addition we recommend reading Venter’s “Testing the Assumptions of Age-to-Age Factors” (Proceedings of the CAS, 1998).


To download updated presentations from the site do a search on the page above on the title “Regression Based Testing of the Assumptions of Underlying Loss Development”

September 2014, San Diego
Limited Attendance Seminar on Reserve Variability


International Congress of Actuaries, April 2014, Washington DC

International Congress of Actuaries March-April 2014, General Information
http://blog.casact.org/2013/07/17/learn-interact-and-grow-at-ica-2014 and http://www.ica2014.org/

Unsupervised Learning Methods Applied to Property Casualty Databases:

A Practical Introduction to Predictive Modeling with Databases (Predictive Modeling Book)


CAS Ratemaking Product Management Seminar
March 2013, Huntington Beach, CA

Predictive Modeling Book – Chapter on Unsupervised Learning


Other (from 2014

The Predictive Modeling Book Project
The Predictive Modeling Book project is a collaborative effort to produce a practical educational resource on predictive modeling for actuaries and others insurance analysts. Many actuaries felt the actuarial exam syllabus and actuarial continuing education were not covering all the essential topics that actuaries involved in advanced analytics and predictive modeling need to be exposed to. In 2011, Ed Frees, Glen Meyers and Richard Derrig, the book’s editors, approached the Casualty Actuarial Society and Canadian Institute of Actuaries for funding and support of a book on predictive modeling for actuaries. Both organizations agreed to support the project.

in addition to the tremendous effort of the editors, graduate students from the University of Wisconsin, where Ed Frees is a professor, made a significant contribution to editing the book. The first volume of Predictive Modeling Applications in Actuarial Science, Volume 1 was published in the fall of 2014. Louise Francis is author of Chapter 12, “Unsupervised Learning.” Volume 2 is in progress. Ms. Francis is authoring a chapter on advanced unsupervised learning for Volume 2.

Casualty Actuarial Society Sponsors Predictive Modeling Book

Amazon Link:Predictive Modeling Applications in Actuarial Science: Vol 1


CAS Limited Attendance Seminar on Reserve Variability
Philadelphia, August 2012

Instructor for 2-day class on data mining with Statsoft
Chicago, June 2012

Salford Analytics Data Mining Conference
Fraud Analysis and Other Applicationsof Unsupervised Learning in P&C Insurance
San Diego, May 2012

Ratemaking and Product Management Seminar
Estimating Systemic Risk for Professional Liability Lines
Philadelphia, March 2012


General Insurance Research Organization Conference
Plenary Session: Overview of Casualty Actuarial Society Research
Workshop on Casualty Actuarial Society Claim Simulator Software
Liverpool, UK, October 2011


CAS Spring Meeting
A Perfect Storm for P&C Analytics
Instructional Approach to Variance Models: Munich Chain Ladder/Capital Allocation
San Diego, May 2010

CAS Seminar on Reinsurance
Financial Crisis – Technical Look Back
New York City, May 2010

CAS Ratemaking and Product Management Seminar
Logic, Fallacies and Paradoxes in Risk Profit Loading
Data Management Call Paper Program: “Text Mining Handbook”
Data Management Call Paper Program: “Data and Disaster: The Role of Data in the Financial Crisis” (winner of paper prize)
Chicago, March 2010


“Introduction to GLMs”, Predictive Modeling Seminar, Casualty Actuary Society.
San Diego, November 2008 Handout

“Predictive Modeling and Insurance Operations”, Casualty Actuary Society Spring Meeting.
Quebec City, November 2008 Handout

“Capitalizing on Decision Trees to Advance Predictive Capabilities.” WRG 2008 Predictive Modeling in Workers Compensation. Handout Texas Closed 7.20.08.xls. Handout

“Text Mining Unstructured Data.” Predictive Modeling Seminar. Casualty Actuary Society. San Diego, November 2008. Handout Work Comp Claims Prediction xls.Handout


“An Introduction to Text Mining with an Application to General Liability Claims Data.” Insightful User Conference 2005. Princeton, NJ. 26-27 October 2005. Handout

“Introduction to Generalized Linear Models.” Special Interest Seminar on Predictive Modeling. Casualty Actuary Society. Chicago, IL. 19 October 2005. Handout

“Trees, Neural Networks and Clustering used for Predictive Modeling.” Spring Meeting. Casualty Actuaries of the Mid-Atlantic. Princeton, NJ. 2 June 2005

“COTOR’s Loss Reserving from the Viewpoint of Modeling.” The Committee on Theory of Risk: Training Project. Casualty Actuary Society Annual Meeting. Phoenix, AZ. 15-18 May 2005.

“Insurance Fraud Detection: MARS vs. Neural Networks” Second Annual CART Conference. Salford Systems. San Francisco, CA. 24 March 2005


“Introduction to Predictive Modeling.” Predictive Modeling Implementation in Workers Compensation. World Research Group. Orlando, FL. 18 December 2004.

“Generalized Linear Models: Introduction.” Panel Presentation. Special Interest Seminar on Predictive Modeling. Casualty Actuarial Society. Chicago, IL. 4-5 October 2004. Download Handout

“The Insolvency Put: An Asset of the Corporation?” Annual Conference. American Risk and Insurance Association. Chicago, IL. 10 August 2004.

“Fraud Classification Using Principal Components Analysis of RIDITS.” Moderator: Panel Discussion. Spring Meeting. Casualty Actuarial Society. Colorado Springs, CO, 16-19 May 2004.

“Data Mining in the Property/Casualty Insurance Industry.” Panel. Ratemaking Seminar. Casualty Actuary Society. Philadelphia, PA. 11-12 March 2004. Download Presentation

“Insurance Fraud Detection: MARS vs. Neural Networks” International CART Data Mining Conference. Salford Systems. San Francisco, CA. March 2004.


“Models- Do You Trust Them? Computer models within the insurance industry.” Panel. Annual Meeting. Casualty Actuarial Society. New Orleans, LA. 9-12 Nov 2003,

“Martian Chronicles: Is MARS Better than Neural Networks?” Paper Presentation. Ratemaking Seminar. Casualty Actuary Society. San Antonio, TX 27-28 March 2003 Download Paper


“Modeling for Fraud: Neural Networks Demystified.” Insurance Fraud Bureau of Massachusetts. November. 2002

“Data Mining Neural Network Applications in Insurance.” Annual Meeting. Casualty Actuary Society. Boston, MA 10-13 November 2002. Download Presentation

“Fair Value Accounting for Actuaries in the Post Enron World.” Loss Reserve Seminar. Casualty Actuary Society. Arlington, VA . 23-24 Sept. 2002.


“Neural Networks Demystified”. Ratemaking Seminar. Casualty Actuarial Society. Las Vegas NV. 11-13 March. 2001

“Fair Valuations Task Force: Methods of determining risk adjusted discounted losses.” Casualty Actuaries of Greater New York. December 2000.

“Fair Value of Insurance Liabilities – CAS White Paper ” Panelist. Casualty Loss Reserve Seminar. Casualty Actuary Society. Minneapolis, MN. 18-19 September 2000.

Presented theory and methods of using regression for estimating loss development factors. ยท. Regression Models of Loss Reserving. September 1998