Adaptive Initialization Method Based on Spatial Local Information for 𝑘-Means Algorithm

Table 1

UCI dataset descriptions (𝑁: number of data points, 𝐷: number of data dimensions, and 𝐾^′: number of classes).

ID	Dataset	𝑁	𝐷	𝐾^′
1	Abalone	4177	7	29
2	Breast cancer Wisconsin (diagnostic)	699	9	2
3	Breast tissue	106	9	6
4	E. coli	336	7	8
5	Glass identification	214	9	6
6	Haberman's survival	306	3	2
7	ILPD (Indian liver patient dataset)	583	10	2
8	Image segmentation	2310	19	7
9	Ionosphere	351	34	2
10	Iris	150	4	4
11	Heart disease (processed Cleveland)	303	13	5
12	ISOLET 5	1559	617	26
13	Letter recognition	20000	16	26
14	Libras movement	360	90	15
15	Lung cancer	32	56	3
16	Multiple features (Fourier)	2000	76	10
17	Musk (version 1, clean 2)	6598	166	2
18	One hundred plant species leaves dataset	1600	64	100
19	Page blocks classification	5473	10	5
20	Parkinson’s disease	195	22	2
21	Pima Indians diabetes	768	8	2
22	Semeion handwritten digit	1593	256	10
23	Spambase	4601	57	2
24	Wine	178	13	3
25	Yeast	1484	8	10