chainladder.model_diagnostics

chainladder.model_diagnostics#

chainladder.model_diagnostics(model: Triangle | MethodBase | Pipeline, name: str | None = None, groupby: str | list(str) | None = None) → Triangle[source]#

A helper function that summarizes various vectors of an IBNR model as columns of a Triangle

Parameters:
model: Triangle|MethodBase|Pipeline

A predicted Triangle, chainladder IBNR estimator or Pipeline

name: str, optional (default=None)

An alias to give the model. This will be added to the index of the return triangle.

groupby:

The index level at which the model should be summarized

Returns:
Triangle with relevant figures as columns, including
  • Latest: Cumulative value at the latest valuation date, equivalent to latest_diagonal
  • Month/Quarter/Year Incremental: Actual emergence between the latest valuation and the one prior valuation date
  • LDF: Age-to-age loss development factor to the next development/valuation period (from ldf_); ignored if groupby is supplied
  • CDF: Cumulative loss development factor from current age to ultimate (from cdf_); ignored if groupby is supplied
  • Ultimate: Projected ultimate loss from the fitted IBNR model (ultimate_)
  • IBNR: Ultimate - Latest
  • Run Off 1/2/3...: Expected incremental emergence in successive future valuation periods (from full_expectation_)
  • Apriori: Expected ultimate for Benktander family of methods (from expectation_)
Columns from the original Triangle are cross-joined into the index. Measure will contain all the columns from the original Triangle.