Bayesian model using KDE

[Filippo Rizzitano]

Hi everyone,

I’m trying to create a Bayesian model using KDE for five phases. In this case, I have intricate relationships (based on stratigraphic data) between the phases, where: Phase 1 overlaps all the other four phases; Phases 2 and 3 overlap each other; Phases 3, 4, and 5 are contiguous.

To address this, I created the following code:

 Plot()
 {
  Phase(Overlap 1)
  {
   Sequence()
   {
    Boundary("Start 1");
    Phase("1")
    {
     KDE_Model("1")
     {
     };
    };
    Boundary("End 1");
   };
   Phase("Overlap 2")
   {
    Sequence()
    {
     Boundary("Start 2");
     Phase("2")
     {
      KDE_Model("2")
      {
      };
     };
     Boundary("End 2");
    };
    Sequence()
    {
     Boundary("Start 3");
     Phase("3")
     {
      KDE_Model("3")
      {
      };
     };
     Boundary("Transition 3/4");
     Phase("4")
     {
      KDE_Model("4")
      {
      };
     };
     Boundary("Transition 4/5");
     Phase("5")
     {
      KDE_Model("5")
      {
      };
     };
     Boundary("End 5");
    };
   };
  };
 };

However, after running this model, OxCal gives me the following warning:
“Warning! Model not supported in this context - KDE_Model.”

I’m wondering if it would be better to use the KDE_Plot command instead, or if perhaps KDE in general isn’t well suited for this type of model with overlapping and contiguous phases?

Thanks in advance for any advice!

Filippo

[Christopher Ramsey]

Yes - if you are using a Bayesian phase model you should use KDE_Plot to show the distribution of events within the phase. This is not part of the model but will generate a KDE distribution of the marginal posteriors within each phase which is I think what you want.

The KDE_Model is used as a standalone model where there is no clear phase structure.

Best wishes

Christopher

> --
> You received this message because you are subscribed to the Google Groups "OxCal" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to oxcal+un...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/51a4aeed-f65c-442d-b879-f43165b54147n%40googlegroups.com.

[Filippo Rizzitano]

Hi Christopher,

Thank you for the clarification! That makes perfect sense now. I’ve switched to using KDE_Plot as you suggested, and it works well without any warnings. It seems to capture the relationships between the phases exactly as I intended.

I appreciate your help in understanding the correct usage of these functions!

Best regards,

Filippo

To view this discussion on the web visit https://groups.google.com/d/msgid/oxcal/302C1394-CFA1-409A-BF9D-A9EDCF0F2F90%40arch.ox.ac.uk.