Ð35Ð
Solving zÈRb for b
¥ Number of
equations = number of time points
H 100s per
run, but perhaps 1000s per subject
¥ Number of
unknowns usually in range 5Ð50
¥ Least
squares solution:
H
denotes an estimate of the true (unknown)
H From , calculate
as the fitted model
o
is the residual time series = noise (we hope)
¥ Collinearity: when matrix canÕt be
inverted
H Near collinearity: when inverse exists but is huge
