TREND and LINEST give different results when applied to the same data
- From: Rune Svendsen <runesvend gmail com>
- To: gnumeric-list gnome org
- Subject: TREND and LINEST give different results when applied to the same data
- Date: Tue, 08 Jun 2010 09:59:46 +0200
I have some data, a set of X- and Y-values, for which I'm trying to
calculate a future Y-value. So far it's going good. It seems the
FORECAST and TREND functions do what I want, which is take all the data
into account, and calculate the future value based on what best fits all
my data points.
FORECAST and TREND produce the same result when I use it on my data.
I have read that TREND basically uses LINEST to calculate its future
value, and as far as I can see, FORECAST is the same as TREND except
that it only calculates a single future value, compared to TREND's
ability to calculate multiple future values.
Now the thing that I find strange is the following. When I apply LINEST
to my data, and thus obtain the slope of the regression line, and the
y-intercept of the line, and use *these* numbers to calculate my desired
future value (which should be how TREND does it, right?), I get a
different future result than I get when I use TREND.
The future values I obtain from TREND and from using the output of
LINEST are the following, respectively: 40407.91963867 and
40407.94565792. Since this number represents a point in time, the
difference amounts to 37 minutes and 28 seconds, which I think is quite
Can anyone explain why I'm seeing this discrepancy?
] [Thread Prev