This ensures that missing values in one of the variable(s) don't propagate, so the combined result will be: 'NA - a', and that the levels of the combined vector are sorted as the levels of the specified variables (levels of the first variable varying first).
interactionWithMissing(data, vars, varSep = " - ")
Vector of length: nrow(data)
,
with interaction vector.
Data.frame with data.
Character vector with variable(s) of interest.
String with separator to which the variable(s) should be combined.
Laure Cougnaud