Reinventing Insurance Modeling: Machine-led GLM in Action

Explore how Radar 5 streamlines GLM model development with machine-led analytics that automate and accelerate the pricing process without sacrificing interpretability or governance.

Building a Generalized Linear Model (GLM) traditionally requires extensive manual effort with data wrangling, variable selection, iterative diagnostics, and validation that can take weeks or months. The webcast "Reinventing Insurance Modeling: Machine-led GLM in Action" demonstrates how Radar 5 transforms this process through machine-led analytics. Using WTW's proprietary Layered GBM, Radar intelligently guides variable discovery, factor selection, and interaction exploration while running hundreds of diagnostics in the background.

The session provides a comprehensive overview of how this approach combines the enhanced intelligence and efficiency of machine learning while maintaining the intuition, explainability, and flexibility of GLM. Through practical demonstrations, you'll see the complete workflow from data import to model validation.

Key Takeaways:

• Automated and Unbiased Modeling: Radar 5's machine-led approach provides a more automated and data-driven methodology for fitting initial GLMs, increasing both efficiency and model quality compared to traditional approaches
• Layered-GBM: The platform uses gradient boosting machine models with sequentially built trees of increasing depth to identify factor importance and interactions, informing intelligent factor selection
• Intelligent Simplification: Built-in algorithms automatically simplify factors into GLM structures with rules for grouping parameters, fitting polynomials, and introducing splines where appropriate
• Seamless Data Integration: Connect directly with Snowflake and DataBricks servers to streamline the data import and export process
• Enhanced Documentation: Export model assumptions and outputs directly into PDF or report formats for easier stakeholder communication and governance

Watch the full webcast to see live demonstrations of the machine-led GLM process in action and discover how Radar 5 delivers faster, more robust model builds within a fully transparent environment.