Some observations on representation of dependency degree k

Dependencies in degree k between attributes have an important role in data mining. To quickly discover these dependencies, speeding up calculation of the dependency degree k is needed. This paper will give some presentations of the dependency degree k. Especially, a new concept of D-stripped quotient set is proposed and by using this concept to presentation of the dependency degree k, the computation time of k will decrease significantly. Also based on the concept of D-stripped quotient set, we present a necessary and sufficient condition for a reduction set of attributes.