generate_transformations_fp: Function to generate all requested FP transformations for a...

In mfp2: Multivariable Fractional Polynomial Models with Extensions

Function to generate all requested FP transformations for a single variable

Description

Function to generate all requested FP transformations for a single variable

Usage

generate_transformations_fp(x, degree, powers)

generate_transformations_acd(x, degree, powers)

Arguments

x	a numeric vector of length nobs assumed to have been shifted and scaled.
degree	numeric indicating the degree of FPs. Assumed to be 2 for acd transformation.
powers	a vector of allowed FP powers.

Details

Any FP transformation is given by a vector of powers, e.g. (p1, p2) for degree 2. These correspond to powers x^p1 and x^p2. Thus, we only need to consider combinations of all values in powers, since order of the entries does not matter. See generate_powers_fp(). A special case are repeated powers, i.e. p1 = p2. In this case, the repeated entries are multiplied by log(x) (see transform_vector_fp()).

When the ACD transformation is requested, then all pairs of length 2 are considered, i.e. 64. See generate_powers_acd().

If degree = 0 then these functions return the data unchanged for fp, or simply the acd transformation of the input variable, i.e. in both cases the power is set to 1 (linear).

Value

A list with two entries:

• data: a list with length equal to the number of possible FPs for the variable of interest. Each entry is a matrix with degree many columns, and nobs observations comprising the FP transformed input variable. For example, for degree = 2 and nobs = 10, each entry is a 10 x 2 matrix. Values are not centered. If degree = 0, the single entry has a single column.

• powers: the associated FP powers for each entry in data.

Functions

• generate_transformations_acd(): Function to generate acd transformations.

mfp2 documentation built on Nov. 15, 2023, 1:06 a.m.