I need Forecasting for excel simplified, specifically forecasting in
regards to decay rate?  Starting from an open ended graph, I need to
predict the decay rate to a certain point, where on the X access it
ends.

---

Request for Question Clarification by jeremymiles-ga on 25 Jul 2002 13:55 PDT

Hello,
Can you tell us a little more about the model you would like to apply
to your data.  Will it be an exponential decay, or some other model.
Would you like 95% confidence intervals as well, or do you just want a
point estimate.

---

Clarification of Question by kstowe1-ga on 25 Jul 2002 14:28 PDT

Sorry, I'm still a little confused myself. :-)  I have a graph that is
updated weekly.  What I'm trying to do is find a way to start at the
specific date the graph ends, predict the weeks ahead of schedule with
the two multipliers we have, until the two new lines drop to intercept
the X axis.  Does that make sense?  I wish I could draw a picture,
it'd be much easier that way!  From the most recent point on the
graph, I need to calculate a decay rate for every week after that with
two different multipliers [giving me two lines -- a split from the
original] and go with them until they hit the X axis.  I'm interested
in the region between the two new lines created.  I'm just not sure
how to graph that?  Thanks again.  Hope this makes sense.

---

Request for Question Clarification by jeremymiles-ga on 25 Jul 2002 16:12 PDT

Sorry, but I am going to burden you with more questions:
Is the line curved or straight?
I am not sure I understand what the two multipliers are - can you
elaborate on this?
Is there a seasonal component to the graph - that is, are there times
of the year when it increases or decreases more or less rapidly.
The standard model for decay is something like radioactive decay,
where the number of particles halves every X units of time.  If this
is the case, the line will never (quite) hit zero.

---

Request for Question Clarification by jeremymiles-ga on 26 Jul 2002 04:56 PDT

If you need more information/assistance than is given in fj-ga's
comment, then request another clarification.  If you don't, I will
assume that no news is good news.

have you tried playing around with the "trend" feature of Excel? From
the Excel help menu:

Project future values and perform regression analysis

You can fill in a series of values that fit a simple linear trend or
an exponential growth trend. The projected values are extrapolated
from starting values that you select on a worksheet. In a linear
series, Microsoft Excel increases or decreases values by a constant
value that is based on the difference between the selected starting
values. In a growth series, Microsoft Excel multiplies values by a
constant value.
To extend complex and nonlinear data, you should use a suitable
worksheet function. The FORECAST function predicts future values based
on a linear regression of a range of known data or known x- and
y-arrays. For example, given corporate earnings for each of the
previous six quarters, the FORECAST function returns earnings expected
for the next two quarters. The similar TREND function and GROWTH
function can extrapolate future y-values that extend a straight line
or exponential curve that best describes the existing data. They can
also return only the y-values based on known x-values for the best-fit
line or curve. To plot a line or curve that describes existing data,
use the existing x-values and y-values returned by the TREND or GROWTH
function. The LINEST and LOGEST functions return various regression
statistics, including the slope and intercept of the best-fit line.

What you want to do is called "extraplation" in mathematics.

My understanding is every week you get a point in your graph. You want
to extrapolate to find out where your decaying graph would end
(intersect X axis (X intercept)).

My guess is best thing for you to do is take a good book on "Numerical
Methods"
Go to the chapter on "Interpolation and Extrapolation". Use one of the
methods that best suits your graph. It would give you the equation of
the curve and even where it would intersect X axis.

To visualise this I would recommend supeimpose the graph created by
this equation over the graph created by your points.

Hope this helps.

OK....here is a very simplified answer....possibly not to your question.

if you have the weeks numbered in col A === row 1 = 1, Row 2 = 2, etc.

And you have ACTUAL data for the first 10 weeks in Col B....

I am assuming that you can graph that.

NOW:  come up with a decay rate, ---- Say .75.

Then in Col B, Row 11 enter +B10*.75

Copy this formula.  It won't ever become 0, but it will get very small.

If this isn't the answer, tell me why.