casact / cascsim Goto Github PK

Request to have a feature of mixed distribution in the frequency or severity distributions.

Current workaround involves using each LSM-defined LOBs as the a distribution to be weighted, and using the frequency of the LOBs as weights. However, doing this loses the ability to model multiple real LOBs.

It would be great if the current LSM could enable mixture distribution feature.

The LSM runs by using an Excel interface file whose name contains "cascsim". In early 2021, the model was running under R 3.x.3. Users changed the R directory to 4.0.4, whereupon the LSM did not run. The original developers, Kailan Shang and Hai You, updated the model to run under R 4.0.4. The name of updated Excel interface contains "cascsim_R4" They did this without charging for their time.

Kailan told me (Joe Marker) that, when a new version of R comes out, there is no guarantee that the existing cascsim file will run under that new version of R. There may be a need to modify the cascsim file. After modifying the model to run under R 4.0.4, Kailan wrote:

"Let us know if that works for you and the RIsk Committee. We have spent 35+ hours on this already and would like to have a formal agreement/contract for further maintenance and enhancement if the CAS is interested. As discussed, many other useful features/models can be added to the existing simulation tool, although the initial scope is defined based on Parodi's paper."

LSM's current parameterization of Pareto is not the more commonly used form of Pareto used in the actuarial community. For example, ISO's parameterization of Pareto could not be directly used in LSM, and must be converted.

It would be better if we could either modify the definition of pareto to Pareto Lomax or add a new distribution in the fitting and simulation process.

Some error notifications are not easy for users to further de-bug.

When model fails to run, users should be able to make some rudimentary clues to diagnose where the root cause of the error. However, the current LSM error often give messages such as "Cannot finish the simulation process properly. Please make sure run control parameters and simulations assumptions are set appropriately."

When running the cascsim file under R 4.0.4, received the following error messages:

Deductible parameter not available
Limit parameter not available

This error disappears when the Deductible le and Limit are both set to "none" in the LoB sheet.

For more detail, see the test documentation file "Test Documentation_PartialSuccessFitting" un der Wiki

Some users have commented that the package does not work and/or underperforms under R version 4.0.x. Downgrading the user R installation to R version 3.6.x appears to resolve the issue.