Conformal prediction for frequency-severity modeling

TL;DR

This paper introduces a model-agnostic conformal prediction framework for insurance claim intervals, applicable to various models including random forests, providing finite-sample guarantees and eliminating calibration set needs.

Contribution

It extends split conformal prediction to two-stage frequency-severity models and leverages out-of-bag data to improve calibration efficiency.

Findings

Effective prediction intervals demonstrated on simulated data.

Real dataset validation confirms practical applicability.

Out-of-bag mechanism reduces calibration data requirements.

Abstract

We present a model-agnostic framework for the construction of prediction intervals of insurance claims, with finite sample statistical guarantees, extending the technique of split conformal prediction to the domain of two-stage frequency-severity modeling. The framework effectiveness is showcased with simulated and real datasets using classical parametric models and contemporary machine learning methods. When the underlying severity model is a random forest, we extend the two-stage split conformal prediction algorithm, showing how the out-of-bag mechanism can be leveraged to eliminate the need for a calibration set in the conformal procedure.