Fig. 9From: Accelerometer sampling requirements for animal behaviour classification and estimation of energy expenditureEvaluation of the influence of normalized window length and normalized sample frequency on the precision and accuracy of the signal amplitude estimation, for simulated sine wave data with and without artificial noise added. A–C The simulated sine wave without noise added (A), with white noise (B), and with the second harmonic added as noise (C). D–I The effect of normalized sampling frequency (f*sampling) and normalized sampling window length (W*sampling) on the signal amplitude estimation. D–F We quantified the estimation precision using the mean of the normalized amplitude difference between signal and estimate (ΔA*), for the sine wave without noise, with white noise, and with the second harmonic, respectively. G–I we quantified the equivalent estimation accuracy using the standard deviation of ΔA*, for the sine wave without noise, with white noise, and with the second harmonic, respectivelyBack to article page