Date:         Tue, 21 Dec 2004 13:00:06 -0800
Reply-To:     cassell.david@EPAMAIL.EPA.GOV
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         "David L. Cassell" <cassell.david@EPAMAIL.EPA.GOV>
Subject:      Re: Somewhat OT statistical programming question
In-Reply-To:  <s1c8413d.022@MAIL.NDRI.ORG>
Content-type: text/plain; charset=US-ASCII

Peter Flom <flom@NDRI.ORG> wrote:
> For those of you who have written statistical programs (in SAS, SAS
IML,
> or otherwise):
>
> How do you estimate the time it will take you to do a project?
>
> e.g., if the project is to create code to run some model that doesn't
> exist in SAS STAT, how do you know if this will take you an hour, a
day,
> a week, or a year?

It depends on whether the 'project' consists of just me writing code
and debugging it, or if it consists of a full-fledged team approach
with meetings to get feedback from end-users.

In case #1, I just sit down estimate time to get the components working
and debug the final process, then (as learned the hard way) apply
Lieutenant Scott's  Rule and up the time to the next level (hours->days,
days->weeks, weeks->months, ...)  Unfortunately, this rule is sometimes
needed.

In case #2, you ought to go with full-fledged Gantt charts and
everything.
Find the bottlenecks.  Identify them publicly.  Hold them up to
company-wide
ridicule when they block your project just because they can.  But do it
all
in a nice way.  :-)

David
--
David Cassell, CSC
Cassell.David@epa.gov
Senior computing specialist
mathematical statistician