More than
Models built by datarobot customers

Insurance Pricing

DataRobot can do claim frequency, severity, and pure premium (loss cost) modeling, helping to avoid being adversely selected against.


P&C Insurance

Why DataRobot

Accuracy in pricing models can save an insurance company a significant amount of money. DataRobot automatically builds models from a wide class of algorithms—from Generalized Linear Models to Random Forests to Deep Learning—assuring that the final model is the best model.

  • Ensure you hit your target loss ratios by avoiding being adversely selected against
  • Push adverse selection onto your competitors
  • Build highly accurate predictive models without data science expertise
  • Integrate real-time predictions in your underwriting system in minutes, without creating business rules and production code


Many P&C insurance companies are seeing deteriorating underwriting results. With the advent of comparative raters in the P&C insurance market, prospects can price compare many companies instantly, often simply choosing the lowest price. Less sophisticated insurance carriers become exposed in the areas where they mispriced to make a sale. The lowest cost may win the business, but may be underpriced relative to the risk. This results in costing a company potentially exorbitant amounts of money in the end.


With DataRobot's wealth of algorithms, you can be confident in the prices you charge. By automating the process of building and comparing models that explore cost versus risk, you can determine whether any risk you consider taking on is priced appropriately. This competitive advantage pushes adverse selection on to competitors, which, overtime will increase your growth and profitability.


  • Avoidance of adverse selection
  • Meet target loss ratios by accurately predicting an insured’s loss costs

Do you have a similar case?

DataRobot did it once before, in the following we will get an even better result and exceed your expectations. Contact us to find out how we can solve your problems.

Contact us