Look to the help file. Search term = trend
line or trend function.
Ross Johnson wrote:
On Sat, 2006-04-01 at 23:35 -0500, Pete Holsberg wrote:
Ross Johnson wrote:
On Sat, 2006-04-01 at 22:27 -0500, Andrew Fisk wrote:
As an idea why don't you do a simple average of the previous 3 months
+ the same month last year (so for April 2006 you would use April
2005 + jan/ feb/mar 2006) and then divide by 4.
Doing more or less the same thing as this, there is the FORECAST
function, which is based on a straight-line best fit. I couldn't find a
function in Calc that does polynomial best fit. If your data shows any
kind of growth that is not a straight line, then you'll need a curve
fitting algorithm.
I'm not quite sure what to do with FORECAST. I have dates (07-2004, 08-2004, etc.) in A and meal counts in B.
I tried =FORECAST(03-2006; B19:B20; A19:A20) to see what I got and it turned out to be #VALUE. I suppose it's unhappy with the dates
If your dates are recognised by Calc as dates (not text) then they
should work ok I believe. The first parameter in FORECAST should be a
cell reference, which should contain the target date.
I found the following link that may help (it's Excel based, but it may
work with Calc too).
http://www.iscpubs.com/articles/abl/b0110bro.pdf
I'll check that out tomorrow. Thanks.
It's overkill I suspect. You need to know the general form of the curve
function at the start by the look of it. But if you can work out how to
plug in a 2nd or 3rd order polynomial, it may do the trick.
Other functions (e.g. LOGEST) are supposed to do similar things, but I
wasn't able to get it to work.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
|
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
