new.pim.poset {pim}

Create a pim.poset environment

Description

This function allows you to create a pim.poset environment that can be added to a pim.environment object. You can use this function to create a custom poset, but in general it's safer to use the relevant arguments of the pim function. That way more safety checks are carried out.

Usage

new.pim.poset(compare, nobs, parent = parent.frame(), ...)

## S4 method for signature 'character,numeric'
new.pim.poset(compare, nobs, parent = parent.frame(), ...)

## S4 method for signature 'matrix,numeric'
new.pim.poset(compare, nobs, parent = parent.frame(), ...)

## S4 method for signature 'list,numeric'
new.pim.poset(compare, nobs, parent, comp.value = "custom", ...)

## S4 method for signature 'matrix,missing'
new.pim.poset(compare, nobs, parent = parent.frame(), ...)

## S4 method for signature 'list,missing'
new.pim.poset(compare, nobs, parent = parent.frame(), ...)

## S4 method for signature 'missing,numeric'
new.pim.poset(compare, nobs, parent = parent.frame(), ...)

Arguments

compare

A character value, matrix or list indicating how the poset should be constructed. Defaults to the default value of create.poset. See Details section for more information.

nobs

An integer value determining the number of observations this poset is created for. If compare is not a character value, the number of observations

parent

An optional environment that serves as the parent for the pim.poset environment. By default this is the environment from which the function is called. Note that for a correct functioning, the parent environment should be set to the pim.environment this object is part of. This is done automatically by the function add.poset.

...

arguments passed to other methods.

comp.value

a character value to be used as value for the compare slot of the object. Defaults to 'custom' and should be left at the default without a very good reason to change it.

Details

A poset (or pseudo observation set) in the context of probabilistic index models is a set of indices that determines which observations are compared with one another. It is used to construct the pseudo-observations on which the model is fitted. You can think of a poset as a "pseudo-observation set".

The most convenient way to use this function, is by specifying a character value for the argument compare. The value "unique" creates a poset in such a way that only unique combinations of two observations are used in the model. The value "all" creates all possible L-R combinations between the observations.

If you want to define the poset yourself, you can pass either a matrix or a list with 2 elements as value for the argument compare. Columns of the matrix or elements of the list should either be named "L" and "R", or be unnamed. When unnamed, the function takes the first column/element as the left poset, and the second column/element as the right poset. If the (col)names are anything else but "L" and "R", these names are ignored and the first column is seen as "L".

Value

an pim.poset object that can be used to replace the poset in a pim environment.

Note

You can omit the argument compare if you supply a value for nobs. You can also omit the argument nobs if you provide a matrix or list as value for compare. The function will try to deduct the number of observations from the highest index value present in the matrix/list

You can't omit both arguments together though, as the function needs at least some information on the number of observations the poset is designed for.

Warning

Changing the value of comp.value by hand might result in errors or a wrongly fitted model. The argument exists for internal purposes and possible extensions later on, but should not be used.

See Also

add.poset for more information on how to adapt the poset of a pim.environment object.

Examples

mypos <- new.pim.poset('unique',n=10) # creates empty environment
ls(mypos)
# Using the created poset functions L and R
# Note this is purely as illustration, this makes no sense
# in the context of a pim analysis.
mypos$L(1:10)
mypos$R(1:10)