Archive of Mr Excel Message Board

Whoa! First of all, dates are linear while your
usage is not. Second, usage is a function of
date -- not vice versa. That's the reason dates
are generally plotted on the x-axis and y=f(x).
And, finally the arguments to TREND() are
(known_y's,known_x's,new_x's,const). There's no
optionalism here.

Thanks. I plotted the data by hand and in accordance
with the basic definition of "method of least squares"
TREND was giving me what it should be. Due to the
current small sample size, the results just look
strange.

As for who is a function of whom, my goal for this
is to supply two values:
1 - What will USAGE be in 4 weeks.
2 - When will USAGE be XXX
Subseuqnetly I need to manipulate both as a function
of the other depending on the context.

The notion of one set of values being linear while
another is not doesn't seem to apply here. (If
both were linear, the data would fit into y=mx+b
pretty easy and I may not need the trend function)

As for TRENDs arguments, the help screen I have
indicate that 'const' will defalut to TRUE if not
supplied. Considering how the function does not
error out on the version of Excel I'm using, I'd
believe the help screen is correct.

Thanks for the responce though. It did get me to
revisit some graphing and the visual clarified the
results I was seeing.

Andy, while I've never had the occasion to use it
you may want to take a look at FORECAST().