E = categorical.empty () returns a $0\times 0$ empty categorical array with no categories. categorical.empty (m, n, …) or categorical.empty (sz) returns an empty categorical array of the requested size, which must have at least one dimension equal to zero. A lone dimension gives a square size, so categorical.empty (3) is an error while categorical.empty (0) is $0\times 0$. As for zeros, a negative dimension counts as zero, and a size vector with nothing in it names no size and gives $0\times 0$.

Source Code: categorical