Description: Predictive modeling is a powerful toolset for assumption setting (mortality, lapse, etc.). The multivariate approach has its advantages in handling high-dimensional data, address correlations, and statistical significance. However, building insurance assumptions is far more than a simple modeling practice. One must take care to address how the model is applied and its business implications. This session will share the learnings from multiple projects.

Description: The insurance industry has a lot to learn from academia's missteps. With predictive analytics assuming an ever-expanding role in risk modeling, carriers would be well-served to establish practices that mitigate the creation of faulty models. Hear practical advice on how to mitigate the risks of overfitting, with a focus on how to translate complex policyholder behavior experience data for long-term products into behavioral models using company- and industry-level data.

Description: Simple mathematical equations often outperform very complex ones in predictive analytics (PA) situations. This session will cover "why" and "how" to guard against overfitting and other potential pitfalls that might not be intuitive.

Description: The article introduces the concept of overfitting in predictive modeling and illustrates how it hinders our ability to predict new data well.

Description: This session will explain how, by applying his or her educational background and business experience, an actuary can quickly learn the basics of data science and perform the important modeling work of insurance.

Description: As any international traveler who has tried to plug in an electronic device knows, the human race is great at devising many different ways to accomplish the same task. Predictive modeling is no different. Data scientists can choose from a wide variety of languages to prepare data, fit models, and verify results. The Rosetta Stone, which showed writing composed in three languages side by side, was instrumental in translating unfamiliar Egyptian hieroglyphics into the more common Greek language. This presentation follows a similar approach by showing the same predictive model implemented in the following languages: R, Python, Spark, MatLab, and SAS. Participants will compare and contrast each of these languages and, by doing so, build a greater understanding of those languages less familiar to them.

Description: In How Significant Is Statistical Significance? The author discusses the p-value that is often used to determine significance. Sometimes we place too much reliance upon canned metrics and lose the ability to notice when something is mathematically sound but not sensible. The article provides three simple examples that require no complicated mathematics to follow yet provide a basis for some deep thought about statistical significance versus substantive significance.