Importance Degree Evaluation of Spare Parts Based on Clustering Algorithm and Back-Propagation Neural Network

Table 4

The Davies–Bouldin (DB) and Calinski–Harabasz (CH) indices of 10 simulations with different k values.

Serial number	k = 2		k = 3		k = 4		k = 5
	DB	CH	DB	CH	DB	CH	DB	CH
1	1.6347	71.6465	0.6859	226.6673	1.1873	122.1868	1.2345	77.5856
2	1.3394	179.4807	0.8963	280.6704	1.1071	165.3401	1.1626	100.0095
3	1.2098	186.3162	0.8821	274.6304	1.2208	112.8648	1.3117	83.3012
4	1.3445	161.4082	1.0001	232.5309	1.2754	132.9948	1.1375	183.1147
5	1.1514	215.2821	0.8887	262.2657	1.2790	154.3030	1.2678	89.4779
6	1.3001	198.2829	0.7225	298.6832	1.1551	174.1512	1.2404	154.8071
7	1.2883	190.4038	0.9551	255.7667	1.1216	126.2768	1.0492	151.5377
8	1.2983	196.1386	1.0411	242.0565	1.1299	140.6781	1.2082	93.3379
9	1.3271	173.8368	0.7084	297.5069	1.0789	214.7179	1.1228	134.8111
10	1.4925	118.0285	0.7278	293.1378	1.1847	133.5798	1.0530	216.0843