As input, bay.ta() expects a matrix of
trait expressions. In its simplest form, this may contain only one
column with a single trait as in the following example from the
Neolithic gallery grave at Sorsum/Germany (“auricular surface”):
data(sorsum_as, package = "baytaAAR")
head(sorsum_as)
#> id auricular_surface
#> 1 700 4
#> 4 704 5
#> 9 710 8
#> 10 711 4
#> 16 717 6
#> 17 719 3In this case, the data are stored as a data.frame. If
this has not been done beforehand, the input is internally converted by
bay.ta() to a matrix. This is necessary even
if – as here – there is only one trait.
Another example is provided with the data from Spitalfields which, among others variables, contains five columns with traits.
data(spitalfields, package = "baytaAAR")
head(spitalfields[,2:6])
#> TO ST MI MA AP
#> 1 1 1 1 1 2
#> 2 2 1 1 1 1
#> 3 1 2 1 1 1
#> 4 3 2 2 1 2
#> 5 1 1 3 1 1
#> 6 2 2 2 1 1NAs are allowed but neither rows nor columns may consist entirely of
NA values. bay.ta will refuse to run in such
cases, and the offending rows or columns need to be removed before
analysis. Please see the
vignette("articles/worked_example") on Chelsea ‘Old church’
for an example of how this can be done.
The levels of all traits must start at 1. Binary traits
are possible. Mixed levels like 1.5 as a short-cut for a
trait expression between 1 and 2, however,
should not be used as this would violate basic principles of ordinal
scaling. Thus, in such cases one of the neighboring levels must be
chosen or they need to be set to NA.
In contrast to what has been published before [@ref_299013], the nodes (i.e., rows of the matrix) do not need to be fully observed for the multinormal model to run, partly because, with NIMBLE vers. 1.4.1, the NIMBLE team introduced a sampler for partially observed multivariate normal variables.