Bias and variance analysis

Bias analysis We assume that the measured unit step response can be represented by 

Proof The bias of the estimated Laguerre coefficient can be computed using Equations (2.72) and (2.75) 

unbiased results for the estimates follow from the condition that = 0. It is also obvious that the estimated coefficients would be biased if (f)] 0. 

Variance analysis We assume that the disturbance (t) has zero mean with autocorrelation function and power spectrum S (w). We also 

---