This function dynamically constructs a regression formula based on provided parameters. It supports incorporating splines for continuous variables and can handle interaction terms, including a subclass variable. The function ensures that the subclass variable is not redundantly included in the baseline variables.
Usage
construct_formula(
Y,
X = 1,
baseline_vars,
continuous_X = FALSE,
splines = FALSE,
subclass = NULL
)Arguments
- Y
A string specifying the dependent variable in the model.
- X
A string specifying the independent treatment or exposure variable; defaults to 1 (intercept only model).
- baseline_vars
A character vector of baseline covariate names to include in the model.
- continuous_X
A logical indicating whether `X` is a continuous variable; if TRUE and `splines` is TRUE, applies spline transformation to `X`.
- splines
A logical indicating whether to apply spline transformations to the treatment variable `X`.
- subclass
An optional string specifying a subclass variable for interaction with `X` and baseline covariates.
Value
A string representing the constructed formula for use in regression modeling functions like `glm`.
Examples
# Example with basic interaction terms without subclass:
construct_formula("health_outcome", "treatment", c("age", "sex"), FALSE, FALSE)
#> [1] "health_outcome ~ treatment * ( age+sex )"
# Example with spline transformation for a continuous treatment:
construct_formula("health_outcome", "treatment", c("age", "sex"), TRUE, TRUE)
#> [1] "health_outcome ~ bs( treatment ) * ( age+sex )"
# Example including a subclass variable:
construct_formula("health_outcome", "treatment", c("age", "sex", "income"), FALSE, FALSE, "region")
#> [1] "health_outcome ~ region * ( treatment * ( age+sex+income ) )"
# Example with continuous treatment, splines, and subclass interaction:
construct_formula("health_outcome", "treatment", c("age", "sex", "income"), TRUE, TRUE, "region")
#> [1] "health_outcome ~ region * ( bs( treatment ) * ( age+sex+income ) )"