Re[4]: Feature suggestion: 2 times for "duration of task": optimistic and pessimistic

On 23 Aug 2002 11:16:48 +0200 Mikael Hallendal <[email protected]> wrote:


MH> From what I understand those three values are needed for the PERT. I
MH> guess they could be entered as you suggest. Perhaps in some way make it
MH> optional which way you'd like to input it?

That would be a solution.
There could be a problem mapping the short date and the long date to a
standard deviation (and back) if they are not equidistant from the target
date. However, i guess that in a first approximation a symmetric
distribution, i.e. the probability of finishing early is the same as the
one of finishing late (Ouch), could be enough.

I guess that using dates kind of disturbs me.
Assume that the target date is 15.9.2002 and that the earliest finishing
date is 8.9.2002. With this information, you can only say that the chance
that the task will be terminated on the 13.9.2002 is the same as the chance
that it will be terminated at the 9.8.2002. This isn't realistic.

Expressing the insecurity of the target date with essentially a probability
distribution, i.e. a deviation, is a much more realistic statement: the
chance of finishing 6 days early is much smaller than the one of
finishing a day early.

Anyway, it doesn't matter too much, as long as things can be expanded
later.

ciao

-- 
Michele Ravani                  [email protected]
"Those who live hoping, die singing" My Gran