Table 2

k-means cluster purity scores (classification accuracies) for varying numbers of clusters (2 through 9) based on row coordinates produced by multiple correspondence analysis

Variable	2	3	4	5	6	7	8	9	Mean score
School	0.97	0.97	0.97	0.97	0.97	0.97	0.97	0.97	0.97
Health	0.94	0.94	0.92	0.91	0.92	0.92	0.92	0.9	0.92
Weapon	0.82	0.82	0.82	0.83	0.83	0.86	0.85	0.87	0.84
Private	0.81	0.81	0.81	0.81	0.81	0.81	0.81	0.83	0.81
Perpetrator	0.48	0.78	0.83	0.92	0.92	0.84	0.88	0.8	0.81
Public	0.78	0.78	0.78	0.78	0.78	0.78	0.78	0.81	0.78
Unknown	0.64	0.65	0.66	0.65	0.65	0.65	0.66	0.84	0.68
Year	0.46	0.58	0.61	0.57	0.57	0.58	0.57	0.64	0.57
Governorate	0.33	0.5	0.52	0.49	0.5	0.49	0.5	0.52	0.48

Eight was the optimal k as identified by the silhouette method.