Analogous Event Loss Set Class

[dwil]

A class designed as a stand in for an Analogous Event analysis.

This class will take a set of loss set sources, event IDs and an occurrence date and build a parametric loss such that the resulting event will only take on the values possible for the supplied event IDs.

Additional Features

• Load: Allows the user to supply a loading factor for the losses.
• Occurrence Probability: Allows the user the specify the frequency of the event (i.e. if the event occurs in 30% of trials)

Additional Notes

• Distributions are reused if possible. This is to avoid duplicate uploads of data.
  • The Severity distribution description includes an MD5 hash of the input data, which is used to identify additional uploads of the same loss data.
  • Frequency and Seasonality distributions will reuse previously created distributions of the same values.
• state_date is used as the date of occurrence for the event
• The class inherits all properties of the parametric loss set.
  • As such it is possible to set the event_id and interpolate properties like any other ParametricLossSet
• An AnalysisProfile is needed for this implementation of the class because it will use filter layers to get the loss values for an specific event ID.
  • This implementation has its limitation but avoids the potential for downloading large data files (i.e. YELTLossSets) and having to parse them client side.