On Wed, 24 Nov 1999, Mr. SISAVATH Sourith wrote:
> I have a data sample of grains and the histogram of the
> grain size makes me think that the distribution is log-normal.
> Is it then reasonable to approximate the density function by a
> log-normal distribution, whose variance and mean value
> has been calculated from the histogram, i.e.
> mean value m = sum [ log x(i)*probability(x(i))]
> variance = sum (log x(i) - m)^2 ???
'Twould seem more straightforward, especially if the raw data be
available in a magnetic medium, to compute the transformed variable
y = log(x)
in your favourite standard statistical package, and calculate the
mean and variance of y. If you need to transform back to the original
metric for some reason, take the antilogs of the mean and variance (or
standard deviation) of y.
This also provides an opportunity for you to eyeball the
distribution of y to see if it's reasonably normal-seeming, thus
helping to justify your first statement above.
> Or is it better trying to fit a log normal distribution using
> methods such as least square method ?
Not sure precisely what you have in mind; but then, perhaps you
aren't sure either...
-- DFB.
------------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
348 Hyde Hall, Plymouth State College, [EMAIL PROTECTED]
MSC #29, Plymouth, NH 03264 603-535-2597
184 Nashua Road, Bedford, NH 03110 603-471-7128
