Methods for detecting seasonal influenza epidemics using a school absenteeism surveillance system

BMC Public Health

Table 5 Best performing statistical models by data type, when missing values are either deleted or linearly interpolated

Data Type	Model	Parameters	FAR	ADD
ES-top
Deleted	Seasonal Mixed, DOW	l = 7-8, Θ = 0.30	0.411	26.71
Interpolated	Seasonal GEE	l = 1, Θ = 0.15	0.350	14.29
ES-3avg
Deleted	Seasonal GEE	l = 15, Θ = 0.25	0.375	29.38
Interpolated	Seasonal GEE	l = 6, Θ = 0.20	0.433	22.75
ES-allavg
Deleted	Seasonal Mixed	l = 11, Θ = 0.25	0.313	23.63
Interpolated	Seasonal Mixed, DOW	l = 7, Θ = 0.20	0.299	15.13
SS-top
Deleted	Seasonal Mixed	l = 4, Θ = 0.10	0.461	14.67
Interpolated	LR, DOW	l = 4, Θ = 0.25	0.454	9.17
SS-3avg
Deleted	Seasonal GEE, DOW	l = 0, Θ = 0.25	0.420	21.00
Interpolated	Seasonal Mixed	l = 1, Θ = 0.15	0.422	21.57
SS-allavg
Deleted	Seasonal GEE, DOW	l = 0, Θ = 0.25	0.420	21.43
Interpolated	Seasonal GEE, DOW	l = 0, Θ = 0.25	0.420	21.43
ES.SS-allavg
Deleted	Seasonal LR	l = 11, Θ = 0.30	0.375	31.75
Interpolated	Seasonal LR	l = 4, Θ = 0.25	0.411	21.86

The metrics for the model with the lowest FAR are shown in bold. See Table 2 for aggregation abbreviations

ISSN: 1471-2458