Re: Average frequency

From: W. Dale Hall (mailtowd_hall_at_pacbell.net)
Date: 08/20/04 

Date: Fri, 20 Aug 2004 19:07:25 GMT

Thomas Magma wrote:

> I have a chart that shows frequency (x) vs. magnitude (y). Lets say there is
> a thousand points on this chart, each point having a different value
> (magnitude). Finding the average magnitude is just the magnitude total over
> the number of points. But how do I find the average frequency? Seeing as
> each frequency is going to have a different weight based on it's own
> magnitude.
>
> Thomas
>
>

I'm sure I don't understand you correctly. Reading your description, I
can't decide whether you're talking about a histogram (sometimes the bin
counts are referred to as frequencies), or the spectrum of some signal
(in which case, the word "frequency" refers to the frequency of some
mode of oscillation, and "magnitude" is an amplitude of that mode of
oscillation).

In either case, I'm not clear about your assertion that the average
magnitude is the magnitude total over the number of points. As I'm
seeing it, magnitudes are always non-negative, and a total of non-
negative values is always greater than or equal to every individual
value, while an average is somehow greater than some values and less
than others.

Perhaps you could state what sort of data it is, what you mean by
the terms frequency and magnitude, and what information you need
to glean from this operation.

Or, you might just take a weighted average of frequencies:

        sum( f_i * magnitude(i))/sum(magnitude(i)),

and see whether that makes any sense (that's my "I don't
know anything, but want to average these things somehow,
so here's a guess and I'll try something else if this turns
out to be just plain stupid" approach).

Dale.