synthesis_model: Define synthesis models

In simongrund1/robosynth: Robust Synthetic Data Generation with Data-Augmented Multiple Imputation

Define synthesis models

Description

Defines a synthesis model for the inclusion in the synthesis.

Usage

synthesis.model(formula, type, fixed = FALSE, proposal = NULL, ...)

Arguments

formula	formula: model formula
type	character: variable type. Possible values are "continuous", "binary", and "categorical".
fixed	logical: if TRUE, the variable is included in the synthesis as a (fixed) covariate without being synthesized.
proposal	(optional) numeric: standard deviation of proposal distribution (only used if type = "continuous"; see Details).
...	(optional): additional arguments passed to the model fitting function (e.g., nnet::multinom for type = "categorical").

Details

This function defines a synthesis model for the inclusion in the synthesis. Each synthesis model includes between and outcome variable and any number of predictors as a formula formula (`outcome ~ 1 + ...`). One or more of these models can be included in the synthesis by passing them to combine.models.

For continuous variables, the proposal argument can be used to specify the standard deviation of the proposal distribution used in the synthesis. If proposal is not specified, combine.models will attempt to set it to a sensible value on the basis of the data.

For usage examples, see combine.models.

Value

An object of class robosynth.synthesis.model.

Author(s)

Simon Grund

See Also

combine.models

simongrund1/robosynth documentation built on March 20, 2022, 6:15 p.m.