Detection of Anomalies in Water Networks by Functional Data Analysis

Table 1

Simulated data with shape outliers. Mean and standard deviation (in brackets) of the percentage of correctly () and falsely () identified outliers over 100 simulation runs.

	= 0	= 0.05		= 0.15
Method
LRT	0 (0)	0 (0)	0 (0)	0 (0)	0 (0)
TRIM	1.3 (0.7)	95.8 (15.1)	0.7 (0.8)	37.4 (35.9)	0.9 (1.2)
POND	1.8 (1.4)	96.6 (11.4)	1.2 (1.1)	2.5 (4.4)	0.2 (0.5)
ISFE	3.0 (2.0)	86.0 (29.7)	2.9 (1.8)	76.1 (33.9)	2.4 (1.8)
RMAH	1.6 (1.5)	97.4 (7.3)	0.9 (1.0)	93.3 (11.9)	0.2 (0.4)
HDR	0 (0)	65.0 (21.0)	1.8 (1.1)	62.9 (13.2)	6.6 (2.3)
FB	0.1 (0.3)	22.2 (21.4)	0.09 (0.3)	9.4 (10.8)	0.4 (0.2)
OUG	4.8 (2.9)	98.6 (5.1)	3.3 (2.2)	86.7 (12.6)	1.0 (1.2)
FOM	0.5 (0.8)	7.6 (15.5)	0.2 (0.5)	0.7 (6.7)	0.6 (2.6)
FOADA	4.8 (2.5)	99.4 (3.4)	3.4 (2.0)	96.2 (10.6)	1.4 (1.3)