These functions allow you to specify the left hand side and
right hand side of a term in a pim model. The user should
only use this functions within a formula using the pim
function. Use in a different context will return an error.
L(x)R(x)
PP(x)
any vector specified in a formula
a vector with the pseudo-observations for x, based on the
poset used to create the function. If used in a wrong context (i.e.
not in a call to pim
), it returns x
unchanged and throws
a warning.
These functions serve only as placeholder. During the fitting process of a pim, they get updated to include the posets (the indices that determine which observations are compared) Note that this makes the functions behave fundamentally different from what you would expect R. The result of these functions depends on the context in which they are called.
These specific functions are actually not used by the
function pim
. pim
calls the internal function
.make.posfun
to create the actual functions L
and R
to work with the specified posets of the model of
interest.
The actual functions used by pim
are saved in a specific environment, a pim.environment
, which resides in the pim-class
object returned by pim
. This way of working
is chosen in order to avoid unnecessary copying of data.
The function PP
serves simply as short for R(x) - L(x)
. If used outside the context of a pim model, it will generate
multiple warnings (see section warning).
# NOT RUN {
pim(income~(L(foodexp) - R(foodexp)), data=Engeldata)
L(1:10) # Gives a warning
# }
Run the code above in your browser using DataLab