This function allows you to generate a weight variable by supplying a set of categorical variables and the target distribution for each of the variables' categories. Weights are computed by iteratively 'raking' conditional 'cells' to the provided marginal targets.
makeWeight(..., name)A series of expressions of the form variable ~ target_weights.
The variable must be a categorical Crunch variable, and the target weights
must be a numeric vector whose length should be equal to the number of
categories contained in the variable, and whose sum is equal to 100 or 1. If
you supply fewer target weights than there are categories makeWeight will
pad the target weight vector with 0s.
The name of the resulting variable
A crunch VariableDefinition() of the weight variable
For instance, if you wanted to create a
weight variable which equally weighted four categories stored in ds$var
you would call ds$weight1 <- makeWeight(ds$var ~ c(25, 25, 25, 25), name = "weight1").
Note that makeWeight returns a VariableDefinition, an expression that
when assigned into your Dataset becomes a derived variable. This does not on
its own set the new variable as "the weight" for your dataset. To set that
attribute, use weight(). Alternatively, you can also create the variable
and set the weight attribute in one step with
weight(ds) <- makeWeight(ds$var ~ c(25, 25, 25, 25), name = "weight1").
weight<-(); settings() for the "default weight" for other
dataset viewers.
if (FALSE) {
mtcars$cyl <- as.factor(mtcars$cyl)
mtcars$gear <- as.factor(mtcars$gear)
ds <- newDataset(mtcars)
# Create a new "raked" variable
ds$weight <- makeWeight(ds$cyl ~ c(30, 30, 40, 0),
ds$gear ~ c(20, 20, 60, 0),
name = "weight"
)
summary(ds$weight)
# ds$weight is not "the weight" for the dataset unless you set it:
weight(ds) <- ds$weight
# Or, you can create the variable and set as weight in one step:
weight(ds) <- makeWeight(ds$var ~ c(25, 25, 25, 25), name = "weight2")
}