Scaled Pearson residual

[Vu1992]

Hi,

In the ODP bootstrap method, i see that we need to calculate (Scaled) Pearson Residual and bootstrap it instead of bootstrap the origin triangle. Can someone show me how to calculate it by this package?

[kennethshsu]

Hi @Vu1992, did you get a chance to follow the tutorial or review the API docs?

I'm not sure what you are trying to do, but it's often not necessary to perform intermediary steps when using an IBNR model.

Please take a look at the links I reference above and let me know if they are helpful or not.

[kennethshsu]

Thank you for stepping in and helping out @jbogaardt! Learning something new everyday!

[Vu1992]

Hi @jbogaardt,
Thank for pointing out the gamma sampling. It mean that when i use IBNR method such as cl.chainladder, it will have the gamma sampling for projected incremental, right?
If it is the case, i think the only suggestion i have is somehow the package allow for manual adjustment to scaled pearson residual because actuaries will do that a lot in practice.

[jbogaardt]

Do you envision something simple as a scale hyperparameter as a direct input in BootstrapODPSample that would be used (if exists) in lieu of calculating from the data?

[Vu1992]

As i understand that scaled hyperparamerter is used for all residuals, I think it will be hard for getting an input like that cause adjustment is usually made for some residuals only, and other will still relied on the calculation from data. I see the package got the drop function in the boostrap, maybe can add something similar like [AY, development,replace residual value]

[Vu1992]

Sorry one more point is that you need to see the Scaled residual triangle to make any adjustment if needed. So that we will need a attribute to show out the scaled pearson residual triangle.