I have a question regarding Gini index: in page 24 lecture 9, it says The attribute
provides the smallest GINIsplit(D) should be chosen. But in the following example, we
didn't compute the GINIsplit(D). Instead we straightly chose the one with lowest GiniA(D),
which is (low, med)(high). Shouldn't we use 0.459 - 0.443/0.458/0.45 to get the smallest
GINIsplit(D), in this case (low, high)(medium)?