Talk:Coherence (statistics)

{{WikiProject banner shell|class=Start|

{{WikiProject Statistics| importance = mid }}

{{WikiProject Mathematics|importance = mid}}

}}

the dimensionality of this expression seems wrong.

should the denominator have squared variables?

also, using the magnitude in the denominator will destroy

any cross-phase information between the two variables.

in signal processing, coherence is defined as:

gxy^2 = ||^2 / [ ]

where Gxy is a cross spectrum and Gxx, Gyy autospectra

thus the magnitude, nor the square of Gxx and Gyy are used.

Random Data 2nd Edition, Bendat & Piersol, John Wiley & Sons, 1986

page 137

::For a moment I thought the above was right. But look closely. The quantity given here is dimensionless. I think that was intended. Michael Hardy 22:27, 9 March 2006 (UTC)