How to forecast?

Hello,

What would be the best data analysis to use for forcasting in the series of numbers below:

July August September October
1,743 5,173 20,330 ?

Thank you in advance

Plotting a quick chart of your three numbers indicates you have exponential growth!
I have assumed that the number of days in a month etc. is not important
so July becomes 1, August becomes 2, September 3, ....

In Cell A1 enter 1, in A2 enter 2, in A3 enter 3 in A4 enter 4.
In Cell B1 enter 1743, in B2 enter 5173, in B3 enter 20330

Now comes the clever bit
In Cell C1 enter =LN(B1); C2 =LN(B2); C3 =LN(B3); then C4 =FORECAST(A4,C1:C3,A1:A3)
Finally in B4 enter =EXP(C4) you should get the answer 66257

LN is the "natural" logarithm and has the nice property that the logarithm of exponential numbers is a simple straight line.
This allows the ordinary FORECAST() function to be employed.
EXP is just the opposite of LN and so brings you back to the original pattern of exponentially increasing numbers from the easy to forecast straight line given by the logarithms.

Mark.

Be Aware
that I have anwered your query from a mathematicians view-point...
In practice exponential growth cannot be sustained for long,
within a very few months the numbers would become ridiculously astronomical!

However given only three numbers it is difficult to suggest a better forecasting option.

Come back when you have 4 numbers...

Mark

Mark,

1st things, thanks for the quick lesson in Logarithms and expotential. I will use in other areas going forward.

I did leave out quit a few variables that would influence the forcast hoping to keep it brief.

While the data apppears to progress rapidly, there will be a number of factors that will alter the data from Oct forward (HDD, CDD,number of days(like you suggested etc.) not to mention previous yrs.

Is there any another option with Forcast()?

Forecast is (arguably) the easiest of the functions available, but it is only linear.
i.e it fits a straight line through the data you give it.
TREND is very similar but it can be made to give a more complex polynomial fit...
(q=a+bx+cy+dz+...)

The method of forecast depends totally on the style of the data.
I would suggest that you use the excel charts,
Find a charting method that subjectively gives you the clearest indication of how the data is changing, and what effects are causing these changes.
You can then look for a forecasting method that matches what you see on the chart.
(this is what I did to decide that your original example was exponential).

The best forecast method is often a pen and paper in conjuction with the human eye!

Mark.