Developing a Clustering-Based Empirical Bayes Analysis Method for Hotspot Identification

Table 3

Characteristics of each component.

Method	Component (sample)	Statistic	Crashes		LW	SW
Time Period 1
-means	Component 1 (527)	Mean	5	10547.19	12.88	10.08	0.54
		SD	6.58	3013.67	1.76	8.45	0.6
	Component 2 (972)	Mean	1.796	4138.07	12.4	9.89	0.55
		SD	2.92	1820.303	1.46	7.77	0.69
Hierarchical	Component 1 (473)	Mean	5.063	10895.39	12.9577	10.24	0.53
		SD	6.58	2988.96	1.803	8.51	0.52
	Component 2 (1026)	Mean	1.939	4314.87	12.39	9.83	0.565
		SD	3.27	1924.28	1.44	7.778	0.72
GFMNB-2	Component 1 (738)	Mean	3	8191	12.58	11.22	0.29
		SD	4.45	3867.52	1.62	8.31	0.17
	Component 2 (761)	Mean	2.85	4646	12.57	8.74	0.81
		SD	5.13	2878.96	1.56	7.53	0.85
Time Period 2
-means	Component 1 (972)	Mean	2.68	4364.14	12.4	9.89	0.55
		SD	4.71	2035.808	1.46	7.77	0.69
	Component 2 (527)	Mean	8.07	11184.04	12.88	10.08	0.54
		SD	10.6	3343.19	1.76	8.459	0.609
Hierarchical	Component 1 (66)	Mean	16.69	18144.65	13.59	12.39	0.64
		SD	17.71	3095.972	2.03	8.46	0.633
	Component 2 (1433)	Mean	4.02	6237.53	12.52	9.84	0.5496
		SD	6.52	3366.45	1.548	7.98	0.66
GFMNB-2	Component 1 (452)	Mean	6.27	9145.69	12.95	12.98	0.26
		SD	8.6	4457.79	1.74	8.12	0.15
	Component 2 (1047)	Mean	3.85	5732.65	12.41	8.66	0.68
		SD	7.33	3546.66	1.49	7.61	0.76