Forecasting Presidential Approval ratings

  • Thread starter Thread starter sfranger2003
  • Start date Start date
S

sfranger2003

This is particularly and exercise, so please no one get offended or heated.
I have a range of years and percentages, but I can't recognize a formula that
will determine the missing values for the months yet to be determined. For
example, Sep 2001 79.7%, Oct 2001 86.4%, Nov 2001 86.2%, and so on. What is
the simplest formula to get the missing values for the months and years to
come?
 
When forecasting the future there is no right answer. If there was I would be
rich by now. There are any number of different algorythms that forecast the
future.

You can try the forecast function but I can't imagine that it will get you
what you want. =forecast(...)

You can do a search on interpolation to see if that helps. The linest
function comes to mind. =linest(...)

You could try Yoda but if I recall he had difficulty seeing the future.
 
sfranger2003 said:
This is particularly and exercise, so please
no one get offended or heated.

I am sure you realize that it is impossible to predict approval ratings, at
least based on past approval ratings alone. It is unfortunate that your
instructor chose such an absurd example to demonstrate specific statistical
concepts.

Deferring to the simplistic thinking that the instructor probably expects....

I have a range of years and percentages,
but I can't recognize a formula that will
determine the missing values for the months
yet to be determined.

The first step in any forecasting effort should be to graph the data or
otherwise determine what model best fits the data -- linear, exponential,
logarithmic, etc.

What is the simplest formula to get the missing
values for the months and years to come?

If you find that a linear model provides the best fit, the following might
be "simplest":

=FORECAST(A25,$B$1:$B$24,$A$1:$A$24)

where A1:A24 has the dates of the known data, B1:B24 has the corresponding
known approval rates, and A25 etc has the unknown dates.

There are other forecast functions for other models.


----- original message -----
 
Back
Top