Description: Actuarial models and modeling processes continue to grow in complexity, accelerated by the changes in U.S. Statutory, U.S. GAAP and IFRS reporting requirements in the past few years. Modeling and valuation teams are expanding their toolkits beyond the traditional vendor-maintained actuarial software and the MS Office suite of tools. Companies have started to use a variety of tools to improve model dev and valuation processes, including: - Data management - Automation and workflow - Reporting and analytics

Description: This is the first in a series of calls for essays on the impact of AI on mortality and longevity. For this call for essays, we are interested in the impact of AI on mortality and longevity modeling, forecasting and prediction.

Description: This is a presentation of recursive APL programs for Steffenson's and Romberg's algorithms. The programs call themselves repeatedly until terminating conditions are reached. From the Actuarial Research Clearing House ARCH 1987 Vol. 1.

Description: The Society of Actuaries (SOA) Research Institute received a grant from the U.S. National Science Foundation (NSF) to educate insurance actuaries on the use of wildfire risk datasets and models to be developed by researchers from academia and the U.S. National Aeronautics and Space Administration (NASA).

Description: This paper reviews the applications of operations research methods in the insurance industry. From Transactions of Society of Actuaries 1995, Vol. 47.

Description: Risk adjustment in government regulated health plans has been around for over two decades. Yet the strategies for promoting risk score accuracy haven't varied much during that time. Most health plans, physician groups, and vendors promote chart reviews and/or in-home assessments as the primary vehicle for collecting data not perceived to be present in the routine claims submission and adjudication stream. We know that most organizations trying to coalesce encounter data from multiple sources often fail to ensure that all relevant data is ultimately submitted to CMS or a state Medicaid agency. Crucial data, often with significant financial value, frequently gets lost in the data pipeline, somewhere between the provider's billing system and the health plan and/or its vendors. Because 837-formatted data is difficult to read, it complicates payers and vendor’s abilities to track encounter data so that every service delivered to patients is credited to the applicable risk adjustment formula. It is incumbent upon successful risk-adjusted entities to be able to track encounter data, identify processes and/or vendors which may be mangling or losing data, in order to ensure that risk scores are sufficiently accurate to be compliant with plan sponsor requirements and to ensure appropriate revenue to the health plan or provider group.

Description: Learn more about the predictive modeling that was applied to industry data for the lapse and mortality experience in the post-level term period. Novel approaches were taken to model the shock lapse, subsequent lapse, and mortality experience. Observe a demonstration of the Tableau tool, which was built to allow users to analyze data on their own.

Description: This paper is from Transactions of Society of Actuaries 1955 Volume 7, Number 17. The author notes that recent advances in calculating equipment make possible producing graduations much more elaborate than any done previously. The paper confines itself to the problem of graduating a single column of rates, such as an ultimate column, using a Whittaker-Henderson B graduation of the fourth order. The author provides a detailed description of the process with significant mathematical theory.

Description: Dive into this review on Artificial Intelligence and Machine Learning (AIML) techniques for actuarial modeling for life & annuities. Presenters will showcase learnings from the recent SOA research paper "Predictive Analytics and Machine Learning – Practical Applications for Actuarial Modeling" (Nested Stochastic). The researchers will go over models and methodologies used to develop high-fidelity proxy models for various life & annuity applications.

Description: Machine learning models are not black-boxes; these models can easily be interpreted using tools like SHAP, LIME and Anchor. These tools can facilitate the understanding of the model and the business problem. We can used these tools to build better models and communicate how predictions are generated. This article demonstrates their use in a mortality model.