Create a forecast estimate updated with actuals weekly

[Steven D'Aprano]

On Fri, 15 Apr 2016 02:36 pm, Sayth Renshaw wrote:

> Hi
> 
> Wondering if someone has this knowledge, and please forgive my maths
> expressions. If I want to estimate need results to achieve a goal at the
> end of a term updated weekly with real results how would I structure this?

I'm not really sure that this question has anything to do with Python. But
then I don't really understand the question -- you haven't explained what
you are trying to do. Remember, it might be clear to you, because you have
been working on the question and understand the background of where is
comes from. But to us it is a big mystery.

 
> So as an example to illustrate my thought process(which could be wrong)
> These are Bills results for the first to weeks.
> 
> Bills Goal = 60% after 5 weeks.

60% of what?


> wk1   wk2     wk3     wk4     wk5
> Bill  54.5%   57.1%

What do these numbers represent? Where do they come from? How do they relate
to the goal of 60%?


> So say these are the results that get it to current.

You've mentioned two weeks. What is "current"?

 
> wk1           wk2
> get   opp     get     opp
> 6     11      4       7
> 
> So 6/11 etc

Where do these numbers come from? What do you mean "get" and "opp"?


> I am thinking to achieve this I then need to estimate an average
> opportunity rate rounded up. (11 + 7)/2 = 9 so if we had a 5 week term
>
> wk3           wk4             wk5
> get   avgopp  get     avgopp  get     avgopp
> X     9       X       9       X       9
> 
> So doing it manually I would discover Bill needs 6 out of 9 each week,
> which results in:
> 
> sumget        28      0.622222222     result
> sumopp        45

Finally something I can understand! 28/45 == 0.62222... but why are you
dividing those numbers? Where do they come from? 3*6/9 == 2, not 28/45.


> But how do I structure this so that the new results when known for week 3
> update and adjust the following estimates?

I don't understand your question. You were able to perform a calculation to
get the result 28/45, so why don't you do the same calculation again (but
this time with three input numbers instead of two) to get the next result?




-- 
Steven