Re: Geo average

From: Myrna Larson (anonymous_at_discussions.microsoft.com)
Date: 01/18/05 

Date: Tue, 18 Jan 2005 16:40:11 -0600

Thanks for the insights. It's a function I haven't used enough to hit the
bugs.

On 17 Jan 2005 16:45:35 -0800, "Harlan Grove" <hrlngrv@aol.com> wrote:

>Myrna Larson wrote...
>>>However, I won't use GEOMEAN for anything
>>
>>Could you explain your reasons?
>
>I use some datasets with large numbers of large or small numbers
>(sometimes I look at X, sometimes at 1/X). These are the sorts of
>datasets which generate overflow or underflow errors when one applies
>the simplistic algorithm for geometric means which Excel's GEOMEAN
>uses. I'm lazy, so I generally stick to one way of doing most tasks so
>they become second nature. In order to avoid this particular problem, I
>always use EXP(AVERAGE(LN(.))), so I don't get nasty surprises when I
>get lazy.
>
>By the same token, I also use DEVSQ(.)/COUNT(.) rather than VARP all
>the time since Jerry W. Lewis clued me into that.
>
>I don't know about you, but if I know some functions have problems in
>extreme cases, I avoid them in all cases. Makes my life easier.