Monitoring

to view this content, you must download the flash player.

Statistics change, and so should your model.

A critical and often overlooked part of a comprehensive predictive analytics initiative is the ongoing monitoring of production data and models. As carriers utilize predictive models to enhance their book of business, and as competitors respond, the data captured and the benefits derived are constantly changing. Carriers need to factor in these transformations in the form of model refreshes to keep their models highly effective and accurate. It is important that carriers not only prepare data initially, but also monitor the data set to routinely check for changes or inaccuracies, assuring that human or technical errors are not resulting in misleading findings. In order to drive accurate decisions, carriers need to recognize the importance of not only cleansing their data for the Modeling & Analytics phase, but also continuing the process throughout the life of the model.

Valen efficiently monitors and manages customer production data and models with the same data validation framework that we use in our Data solutions. This allows us to quickly and easily refresh models, and ensures that our customers' models retain their expected levels of performance without additional customer investment in IT systems and resources.

Valen also provides reporting capabilities as part of the monitoring process. Below is a sample list of some of Valen's standard reports:

Risk factor visualizations
Model analyzer report
Model lift visualization
Model version comparison

Valen periodically develops and provides additional standard reports based on customer feedback and market requirements.

Valen's customers experience tremendous value by closing the loop with Valen's Monitoring & Reporting capabilities. This step is often overlooked, or at a minimum underestimated, and poorly executed by carriers unfamiliar with some of the pitfalls of predictive analytics initiatives. Valen provides peace of mind that if a customer successfully navigates the data, analytics and deployment phases, the customer will not fall down by losing control of a production model - a potentially costly mistake.