A Log-Based Anomaly Detection Method with Efficient Neighbor Searching and Automatic K Neighbor Selection

Table 6

Comparison of the overall searching time(s).

Dimension		30	100	300	100
Data size		1 G			10 M	100 M	1 G
Liberty	Our method	12.38	27.34	35.02	0.281	2.243	27.34
	Traditional kNN	78.39	171.98	336.36	2.63	21.05	171.98
BGL	Our method	11.09	21.52	33.42	0.279	2.546	21.52
	Traditional kNN	35.01	103.35	198.46	1.68	13.18	103.35
Thunderbird	Our method	13.71	28.25	29.34	0.273	2.037	28.25
	Traditional kNN	114.68	236.26	440.97	3.2	27.29	236.26
Spirit	Our method	12.63	30.28	34.58	0.254	1.899	30.28
	Traditional kNN	112.21	247.79	463.87	3.14	27.33	247.79
HDFS	Our method	6.16	19.79	23.15	0.181	1.32	19.79
	Traditional kNN	82.08	177.69	320.46	2.51	20.99	177.69
Zookeeper	Our method	8.01	17.49	21.76	0.224	1.68	17.49
	Traditional kNN	86.29	197.25	366.99	2.79	23.5	197.25