I would like to know if there is a solution for computing MD (mean
deviation), which would allow me to compute it in one pass.  i.e., w/o
having to compute <x> the mean first.

MD = 1/(N-1) * Sum(i=1 to N)||(xi - <x>)||


I can do what I'm asking for for the case of SD (standard deviation)

SD = sqrt( 1/(N-1) * Sum(i=1 to N)(xi - <x>)^2 )

expanding and using the fact that N * <x> = Sum(i=1 to N)(xi)

SD = sqrt( 1/(N-1) * Sum(i=1 to N)(xi^2)  -  (1/(N-1) * Sum(i=1 to
N)(xi))^2  )

but unfortunately the same trick does not work w/ the absolute value
in the MD.

I already consulted 
http://mathworld.wolfram.com/MeanDeviation.html
w/o any real help.

Thanks in advance,
Michael
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to