2023 Virtual Predictive Analytics Bootcamp

Event Details

-
Wednesdays and Fridays 12:00 PM – 2:30 PM (ET)

Microsoft Teams

About This Event

The virtual Predictive Analytics (PA) bootcamp is a six-week virtual course.

The PA Bootcamp will cover topics from both the classical statistics and machine learning paradigms.  The content will be self-contained and designed for beginners with little or no experience with data science. We will leverage and provide introductions to predictive modeling in R and Python.

The content will be delivered using a blend of: (i) asynchronous virtual content to set up each session, (ii) live virtual lectures with breakout sessions that communicate, demonstrate, and discuss the key concepts and techniques; and (iii) self-lead interactive case studies that further dive into application of the techniques and concepts, and cement the content.

The interactive case studies will be provided in codebooks using relevant insurance datasets. These interactive exercises will serve as a “bridge” between theoretical discussions and practical applications. The exercises are open-ended so that learners can explore and further develop the applications.

Time Commitment: There are two 2.5 hour sessions each week during the six-week period (Wednesdays and Fridays), during which the instructors introduce and illustrate the relevant concepts and techniques, students discuss various aspects in smaller break-out sessions, and introduce a detailed case study that shows the application of the techniques in an actuarial application is introduced. Learners are expected to watch a few short videos and read one to two articles to prepare for each session (about 30 minutes of pre-work). Furthermore, students are expected to work through a self-guided tutorial following each session that showcases the material in a hands-on fashion and they can further explore on their own (about 2 hours of review and application). Therefore, the total bootcamp will require about 50 hours of learners’ time over the ten weeks.

INITIAL PREWORK: Please note that prework videos will be provided two-weeks prior to the start of the bootcamp via the Predictive Modeling Bootcamp Community.

  • R and R Studio Installation (for R Programming)
  • Set up Google Colab Work environment (for Python Programming)
  • Video 1: Data Basics
  • Video 2: Data Fundamentals and Elementary Statistics in R
  • Video 3: Data Fundamentals and Elementary Statistics in Py-Pandas

 

Attendance is limited to 35 participants, individual registrations only. Group registrations are not permitted.

Event Information

Casualty Actuarial Society's Envisioned Future 

The CAS will be recognized globally as the premier organization in advancing the practice and application of casualty actuarial science and educating professionals in general insurance, including property-casualty and similar risk exposure. 

Continuing Education Credits 

The CAS Continuing Education Policy applies to all ACAS and FCAS members who provide actuarial services. Actuarial services are defined in the CAS Code of Professional Conduct as "professional services provided to a Principal by an individual acting in the capacity of an actuary. Such services include the rendering of advice, recommendations, findings or opinions based upon actuarial considerations". Members who are or could be subject to the continuing education requirements of a national actuarial organization can meet the requirements of the CAS Continuing Education Policy by satisfying the continuing education requirements established by a national actuarial organization recognized by the Policy. 

This activity may qualify for up to 30 CE credits for the 2023 Virtual Predictive Modeling Bootcamp for CAS members. Participants should claim credit commensurate with the extent of their participation in the activity. CAS members earn 1 CE credit per 50 minutes of educational session time, not to include breaks or lunch. 

Note: The amount of CE credit that can be earned for participating in this activity must be assessed by the individual attendee. It also may be different for individuals who are subject to the requirements of organizations other than the American Academy of Actuaries. 

Virtual Bootcamp Recordings 

Recordings of this bootcamp will be available to attendees on the Predictive Modeling Bootcamp Community for five years.  

Speaker Opinions 

The opinions expressed by speakers at this event are their own and do not necessarily reflect the opinions of the CAS. 
  
Contact Us  

For more information on content, please contact Wendy Ponce, Professional Education Coordinator, at wponce@casact.org.  

For more information on course logistics or attendee registration, please contact Leanne Wieczorek, Meeting Planner at lwieczorek@casact.org.  

For more information on other CAS opportunities or regarding administrative policies such as complaints and refunds, please contact the CAS Office at (703) 276-3100 or office@casact.org

Registration Information

Register

Limit Up to 35 Participants

Group Registrations are not Permitted

Due to required pre-work, registration for this event will close on August 30, 2022.

 

 On or Before August 16, 2023

Fee After August 16, 2023

MEMBER/ICAS/Subscriber/Candidate

$1,350

$1,450

NON- MEMBER 

$1,550

$1,650

Cancellations/Refunds

Registrations fees will be refunded for cancellations received in writing at the CAS Office via email, refund@casact.org, by August 24, 2023, less a $200 processing fee. 

Speakers

Dani Bauer is a Professor and the Hickman-Larson Chair in Actuarial Science in the Department of Risk and Insurance of the Wisconsin School of Business, University of Wisconsin-Madison.  Dani specializes in the development of models for the valuation and risk management of insurance products and insurance-linked securities.  His research publishes in leading journals in actuarial science, finance, management, and statistics, and he serves on the editorial boards of several journals in actuarial science and risk management.  Dani teaches classes in actuarial science, quantitative finance, and data analytics. He is one of the architects and currently serves as the director of the master’s in Business Analytics at the Wisconsin School of Business. Dani received his doctorate in Mathematics from Ulm University, Germany, from where he also holds a Diploma in Mathematics and Economics.  Furthermore, he obtained an M.S. degree from San Diego State University, where he studied Statistics as a Fulbright scholar.

Jim Guszcza is a Research Affiliate at Stanford’s Center for Advanced Study in the Behavioral Sciences, leading a project to articulate a practical field of human-machine hybrid intelligence.  Jim worked as data scientist for over two decades and was the first person to be designated Deloitte’s U.S. Chief Data Scientist. Jim holds a PhD in Philosophy from The University of Chicago, is a Fellow of the Casualty Actuarial Society, and spent time as a professor at the University of Wisconsin-Madison’s Department of Risk and Insurance. Jim serves on the advisory boards of the Psychology of Technology Institute and the Penn Center for Social Norms and Behavioral Dynamics.

Peng Shi is on the faculty of the Risk and Insurance Department at the University of Wisconsin-Madison. He is also the Charles and Laura Albright Professor in Business and Finance. Peng is an Associate of the Casualty Actuarial Society (ACAS) and a Fellow of the Society of Actuaries (FSA). He teaches classes in actuarial science and machine learning at undergraduate level, and longitudinal and panel data analysis at graduate level. His research interests are at the intersection of insurance and statistics. He has won various research awards in actuarial science, including the Charles A Hachemeister Prize, American Risk and Insurance Association Prize, Ronald Bornhuetter Loss Reserve Prize, and IAA Best Paper etc.  He also serves on the editorial board of several scholarly journals in actuarial science. Peng holds a Ph.D. in actuarial science, risk management, and insurance with a minor in economics from the University of Wisconsin-Madison

Schedule

INITIAL PREWORK:

  • R and R Studio Installation (for R Programming)
  • Set up Google Colab Work environment (for Python Programming)
  • Video 1: Data Basics
  • Video 2: Data Fundamentals and Elementary Statistics in R
  • Video 3: Data Fundamentals and Elementary Statistics in Py-Pandas

 

Session

Date

Time

Topics Covered

1

Sep 6

12:00 PM – 2:30 PM (ET)

Review, Some Fundamental Techniques, and Overview

  • Overview
  • Univariate statistics
  • Bootstrapping
  • Multivariate statistics, correlation
  • OLS Regression
  • Prediction vs. Causal inference (Correlation vs. Causation)

Prework: Introductory quiz on student background
Tutorial: Bootstrapping confidence intervals of risk measures using fire loss data

2

Sep 8

12:00 PM – 2:30 PM (ET)

 Elementary Generalized Linear Models (GLMs)

  • From OLS to GLM
  • Gamma Regression
  • Poisson Regression

Prework: GLM video, quiz
Tutorial: Predicting auto collision losses (insuranceData, DeJong Heller Diabetes data)

3

Sep 15

12:00 PM – 2:30 PM (ET)

 Advanced Generalized Linear Models (GLMs)

  • Frequency-Severity Modeling
  • Tweedie Model
  • In-sample vs. out-of-sample error, AIC
  • Building GLMs Best Practices
  • Advanced Actuarial Applications (longitudinal modeling, probabilistic forecasting, etc.)

Prework: Scholarly article, quiz
Tutorial: Predicting claims in auto insurance, revisited (using French Auto data freMTPL)

4

Sep 20

12:00 PM – 2:30 PM (ET)

Regression in the 21st Century: Regularized Regression and LASSO

  • In-sample and out-of-sample error revisited, optimism
  • The bias-variance tradeoff
  • Ridge and Lasso Regression, Regularization
  • Cross-Validation
  • Penalized GLM

Prework: Video on optimism
Tutorial: Predicting claims in auto insurance, revisited (Allstats loss data)

5

Sep 22

12:00 PM – 2:30 PM (ET)

Non-Linear Modeling

  • Modeling Non-linearities
  • From Polynomial Regression to Regression splines
  • Kernel regression
  • Generalized Additive Models

Prework: Set of videos introducing non-linear modeling techniques
Tutorial: Predicting claims in auto insurance, revisited (air quality dataset)

6

Sep 27

12:00 PM – 2:30 PM (ET)

Introduction to AI and Machine Learning

  • Michael Jordan ML and Breiman’s two cultures
  • Use cases for ML, predicting future and automation
  • k-nearest-neighbors and support vector machines
  • Regression and Classification
  • MAPE, MSE, Confusion matrix, ROC-curves, AUC

Prework: Articles by Michael Jordan and Leo Breiman
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality)

7

Sep 29

12:00 PM – 2:30 PM (ET)

Tree-Based Models, Bagging and Boosting

  • Classification and Regression Trees (CART)
  • Bootstrap Aggregation and Random Forests
  • Boosting, Boosted Trees

Prework: Video on tree-based modeling
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality); revisiting insurance examples

8

Oct 4

12:00 PM – 2:30 PM (ET)

Neural Networks and Deep Learning

  • Perceptron, regression and trees
  • Artificial Neural Networks
  • Deep Learning
  • Combined Actuarial Neural Networks

Prework: Video and quiz
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality); revisiting insurance examples

9

Oct 6

12:00 PM – 2:30 PM (ET)

ML Applications in insurance—practical and regulatory challenges

  • Challenges: Black-box character, regulation, bias and fairness
  • Explainable AI
  • “Fair” algorithms, criteria and mechanisms
  • Trading off

Prework: Articles on algorithmic bias and insurance
Tutorial: Auto dataset with protected characteristics, trading off accuracy and fairness

10

Oct 11

12:00 PM – 2:30 PM (ET)

Beyond Prediction and Outlook

  • Limitations of AI, AI Snake oil
  • Algorithmic decision making
  • Wrap-up

Prework: Articles on AI snake oil and human machine interaction
Tutorial: --

 

Schedule

INITIAL PREWORK:

  • R and R Studio Installation (for R Programming)
  • Set up Google Colab Work environment (for Python Programming)
  • Video 1: Data Basics
  • Video 2: Data Fundamentals and Elementary Statistics in R
  • Video 3: Data Fundamentals and Elementary Statistics in Py-Pandas

 

Session

Date

Time

Topics Covered

1

Sep 6

12:00 PM – 2:30 PM (ET)

Review, Some Fundamental Techniques, and Overview

  • Overview
  • Univariate statistics
  • Bootstrapping
  • Multivariate statistics, correlation
  • OLS Regression
  • Prediction vs. Causal inference (Correlation vs. Causation)

Prework: Introductory quiz on student background
Tutorial: Bootstrapping confidence intervals of risk measures using fire loss data

2

Sep 8

12:00 PM – 2:30 PM (ET)

 Elementary Generalized Linear Models (GLMs)

  • From OLS to GLM
  • Gamma Regression
  • Poisson Regression

Prework: GLM video, quiz
Tutorial: Predicting auto collision losses (insuranceData, DeJong Heller Diabetes data)

3

Sep 15

12:00 PM – 2:30 PM (ET)

 Advanced Generalized Linear Models (GLMs)

  • Frequency-Severity Modeling
  • Tweedie Model
  • In-sample vs. out-of-sample error, AIC
  • Building GLMs Best Practices
  • Advanced Actuarial Applications (longitudinal modeling, probabilistic forecasting, etc.)

Prework: Scholarly article, quiz
Tutorial: Predicting claims in auto insurance, revisited (using French Auto data freMTPL)

4

Sep 20

12:00 PM – 2:30 PM (ET)

Regression in the 21st Century: Regularized Regression and LASSO

  • In-sample and out-of-sample error revisited, optimism
  • The bias-variance tradeoff
  • Ridge and Lasso Regression, Regularization
  • Cross-Validation
  • Penalized GLM

Prework: Video on optimism
Tutorial: Predicting claims in auto insurance, revisited (Allstats loss data)

5

Sep 22

12:00 PM – 2:30 PM (ET)

Non-Linear Modeling

  • Modeling Non-linearities
  • From Polynomial Regression to Regression splines
  • Kernel regression
  • Generalized Additive Models

Prework: Set of videos introducing non-linear modeling techniques
Tutorial: Predicting claims in auto insurance, revisited (air quality dataset)

6

Sep 27

12:00 PM – 2:30 PM (ET)

Introduction to AI and Machine Learning

  • Michael Jordan ML and Breiman’s two cultures
  • Use cases for ML, predicting future and automation
  • k-nearest-neighbors and support vector machines
  • Regression and Classification
  • MAPE, MSE, Confusion matrix, ROC-curves, AUC

Prework: Articles by Michael Jordan and Leo Breiman
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality)

7

Sep 29

12:00 PM – 2:30 PM (ET)

Tree-Based Models, Bagging and Boosting

  • Classification and Regression Trees (CART)
  • Bootstrap Aggregation and Random Forests
  • Boosting, Boosted Trees

Prework: Video on tree-based modeling
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality); revisiting insurance examples

8

Oct 4

12:00 PM – 2:30 PM (ET)

Neural Networks and Deep Learning

  • Perceptron, regression and trees
  • Artificial Neural Networks
  • Deep Learning
  • Combined Actuarial Neural Networks

Prework: Video and quiz
Tutorial: Classification tasks (credit card defaults, insurance sales, wine quality); revisiting insurance examples

9

Oct 6

12:00 PM – 2:30 PM (ET)

ML Applications in insurance—practical and regulatory challenges

  • Challenges: Black-box character, regulation, bias and fairness
  • Explainable AI
  • “Fair” algorithms, criteria and mechanisms
  • Trading off

Prework: Articles on algorithmic bias and insurance
Tutorial: Auto dataset with protected characteristics, trading off accuracy and fairness

10

Oct 11

12:00 PM – 2:30 PM (ET)

Beyond Prediction and Outlook

  • Limitations of AI, AI Snake oil
  • Algorithmic decision making
  • Wrap-up

Prework: Articles on AI snake oil and human machine interaction
Tutorial: --