Improved Density Peaks Clustering Based on Natural Neighbor Expanded Group

Algorithm 1

DPC-NNEG.

	Require: Dataset , the goal number of clusters G
	Ensure: The result of clustering:
(1)	Create a k-d tree;
(2)	Search the k-d tree;
(3)	Determine NN according to [34], and record , which means determined;
(4)	Calculate local density according to equation (8).;
(5)	Assign each point to its NDP of its NNE to generate several NNEGs;
(6)	Create a matrix AGG = (, );
(7)	for i = 1 : n do
(8)	for t = 1 : do
(9)	if the tth NNE and sample i belongs to different NNEGs do
(10)	Calculate the closeness degree of this edge, referring to equation (10);
(11)	Add the DC of this edge to the corresponding unit of AGG;
(12)	end if
(13)	end for
(14)	end for
(15)	while the number of clusters does not equal G do
(16)	Store zero in the unit with the min value but greater than zero;
(17)	Count the number of clusters;
(18)	end while