[comp.software-eng] Info wanted on Guidelines for Metrics

seb1@druhi.ATT.COM (Sharon Badian) (02/19/91)

I am looking for pointers to any information on creating a good
metrics program on the project level. I am looking for guidelines
for a project manager if he/she wants to embark on a program to
collect in-process metrics. I have heard of Basili's goal-directed
metrics paradigm, but I can't seem to find any references to it.
Anyone know if he actually wrote a paper on this? I heard about it
at a talk he gave at AT&T Bell Labs about 3 years ago.

Also, if you have instituted a sucessful metrics program for your
project, I'd be interested in hearing from you. I'm less concerned
about the actual metrics you collected and more interested in how
you determined which metrics you would collect and how you went
about getting the data.

Thanks for any help!!
Sharon Badian
att!druhi!seb1

fjs@bonz.enet.dec.com (Jay Shields) (02/21/91)

A good book on setting up and collecting metrics is "Controlling
Software Projects: Management, Measurement & Estimation" 
Prentice-Hall, 1982  by Tom DeMarco. 

   DeMarco also has a project-level metrics tool called Checkpoint. 
I know it runs on DOS, but am not sure if any other OS's are
supported.
You can contact his company at:

   Software Productivity Research, Inc.
   77 South Bedford Street
   Burlington, MA 01803
   (617) 273-0140

   But be forwarned: the tool will cost you a bundle.

   Hope that helps!

--
F. Jay Shields
Digital Equipment Corporation
Littleton, MA
fjs@bonz.ogo.dec.com
(508) 952-3238

warren@eecs.cs.pdx.edu (Warren Harrison) (02/21/91)

In article <7574@drutx.ATT.COM> seb1@druhi.ATT.COM (Sharon Badian) writes:
>I am looking for pointers to any information on creating a good
>metrics program on the project level. I am looking for guidelines
>for a project manager if he/she wants to embark on a program to
>collect in-process metrics.
.
.
.
>Also, if you have instituted a sucessful metrics program for your
>project, I'd be interested in hearing from you. I'm less concerned
>about the actual metrics you collected and more interested in how
>you determined which metrics you would collect and how you went
>about getting the data.
>

In about a month (March 17-19) we'll be hosting the Third Annual Oregon
Workshop on Software Metrics. There will be at least six presentations
that will focus on this topic (plus well over a dozen other presentations that
address other metrics issues):

"Implementing a Metrics Program", Steve Wilkinson of ACUCOBOL

"Using Process Metrics with an Unstable Process", Mike Perdue of Sun Microsystems

"An Application of Applying Metrics to the Software Review Process", Carl
Seddio of Eastman-Kodak

"Experience With Complexity Metrics for Coding Standards", George Stark of
the MITRE COrporation

"U.S. Army Test & Evaluation Panel Software Metrics", Janet Beavers, US Army

"Software Metrics Reporting: Presentation of Multiple Metrics for Analysis
of Improvement", Sherri Lawrence-Pfleeger, Contel Technology


Note that each of these presentations are being made by industrial folks
who have been actually applying them (as opposed to us academic types who
actually think about them :-). The mix of presentations is  12 industry/10
academic, so you get to think about metrics too ...

We'll also have a tutorial each evening, and a panel session on metric
standardization and a panel session on getting senior management to use
metrics.

- blatant commercialism on -
Now, what would such an information packed Workshop normally cost? $700?
$800? $900? a $1,000? Well ... have we got a deal for all of you out there,
the cost of admission here is only $525 ... but wait! there's more! Unlike
most other workshop/conferences we don't nickle and dime you to death with
a meal here and a $150 hotel room there ... for your $525 you get (gasp)
all meals, lodging, tutorials, and transportation from the airport (of course
the shuttle back costs $1,500 ... just kidding'). But hurry! We're limiting
it to 75 attendees, and we've already filling up, so give us a call at
503-690-1460 and tell them you want to sign up for AOWSM '91.
- blatant commercialism off -

Hope to see all of you in March :-)

Warren



==========================================================================
Warren Harrison                                          warren@cs.pdx.edu
Center for Software Quality Research                          503/725-3108
Portland State University/CMPS   

seb1@druhi.ATT.COM (Sharon Badian) (02/21/91)

in article <1991Feb20.180831.28131@e2big.mko.dec.com>, fjs@bonz.enet.dec.com (Jay Shields) says:
>    DeMarco also has a project-level metrics tool called Checkpoint. 
> I know it runs on DOS, but am not sure if any other OS's are
> supported.

Check your info! DeMarco does not produce CHECKPOINT. Capers Jones does.
We already own it. CHECKPOINT is not exactly a metrics tool though you
can use it as such. It's a cost/schedule estimation tool.

Sharon Badian
att!druhi!seb1

locke@paradyne.com (Richard Locke) (02/22/91)

In article <1991Feb20.180831.28131@e2big.mko.dec.com> fjs@bonz.enet.dec.com (Jay Shields) writes:

>   DeMarco also has a project-level metrics tool called Checkpoint. 
>I know it runs on DOS, but am not sure if any other OS's are
>supported.

It's not DeMarco, but Capers Jones who's behind Checkpoint.

Checkpoint just helps you keep track of some metric & milestone
information (as opposed helping you decide what metrics to track, which
is very different).  Its substantial price tag is justified more
(in my opinion) in its predictive abilities than its date storage
abilities...  Of course, I'm a developer with no direct need to give
upper management nice impressive graphs, etc ;-)

Addressing the original issue:  "Software Metrics: Establishing
a Company-Wide Program" by Grady & Caswell has some practical
stuff in it, though it obviously is focused less on the project
level than the company level.  I think you'll find it useful
nonetheless.

By the way, I've been working to evaluate project estimation tools
lately which is why I'm familiar with Checkpoint.  I'd be interested
in comparing notes with others who've done work in this area recently.


-dick

{uunet,peora}!pdn!locke			Phone: (813) 530-8241

seb1@druhi.ATT.COM (Sharon Badian) (02/25/91)

in article <1991Feb22.163103.26489@e2big.mko.dec.com>, fjs@bonz.enet.dec.com (Jay Shields) says:
> As for CHECKPOINT's application to the question at hand,  I thought
> that
> the request for information about starting a metrics program at the 
> project level was sufficiently general that CHECKPOINT would apply. 
> It's been my experience that starting a full-blown metrics program
> from
> scratch can be rather difficult; tools such as CHECKPOINT often form
> the 
> initial phase of such a program. It would be interesting to hear from
> anyone who uses tools such as CHECKPOINT as part of their metrics
> program
> (or from those who think it has no such place). 

I was the person who made the initial request and the person who has
CHECKPOINT. So, I can answer the question of CHECKPOINT as a metrics
tool. I don't find it too helpful because there is no focus. You can
collect tons of information and not be too sure what you are learning
about your process. It is overwhelming. I want a smaller set of metrics
that allow me to answer important questions about my software process,
not a million of them that tell me everything I would ever want to know
(including things I don't really care about). That's why I'm interested
in the Basili goal/question/metric paradigm. It links metrics directly
to things you care about.

Sharon Badian
att!druhi!seb1