A Robust k-Means Clustering Algorithm Based on Observation Point Mechanism

Table 1

Comparison between k-means clustering algorithm and our proposed clustering algorithm.

Data sets	N	t (%)	k	d
Synthetic #1	84	2.3	2	2	75	27	0.084	0.954	0.020	0.000
Synthetic #2	236	4.2	7	2	80	107	0.791	0.860	0.063	0.016
Synthetic #3	2557	2.2	5	3	92	761	0.774	0.972	0.047	0.031
Synthetic #4	3670	1.9	6	4	96	656	0.815	0.977	0.188	0.063
Synthetic #5	3655	1.5	6	5	96	573	0.816	0.982	0.313	0.078
Synthetic #6	2830	1.1	7	6	95	296	0.848	0.988	0.250	0.063
Iris	150	0	3	4	80	27	0.730	0.730	0.031	0.016
Iris	152	1.3	3	4	82	26	0.531	0.743	0.031	0.016
Seeds	210	0	3	7	80	37	0.717	0.728	0.047	0.016
Seeds	212	0.9	3	7	80	37	0.462	0.694	0.047	0.016
Wine	178	0	3	13	81	6	0.870	0.850	0.031	0.016
Wine	180	1.1	3	13	81	10	0.365	0.882	0.031	0.016

The real data set which includes two synthetic outliers as shown in our BaiduPan.