|We can then
divide the dataset into two partitions.
in D(i), if not null or constant, is correlated with one or more attributes
|We can consider
only the attributes in D(r) to perform FC.