Dear Splus gurus,
I am often in the situation that I am given an experiment to analyze
that was (heaven forbid!) designed, ummmm, less than carefully.
The first step is often to determine exactly which conditions are
"comparable" and then to gradually determine which sets can be grouped
together to form statistically useful experiments (factorial-type
groups, essentially).
Of course, being lazy, I would really like to automate this task of
analyzing the design table. It seems to me that this should be a
well-known graph theory problem since a factorial experiment constitutes
a closed graph and any two points that are connected are comparable.
"connection" being defined as differing in only condition.
Of course, this doesn't allow me to find such things as OA's in the
design table, but it is at least a start.
I can certainly write some scripts to (rudimentarily) do this task.
However, since it seems like a graph problem, someone out there may have
already implemented something similar. Alternatively, is there a way to
trick any of the existing clustering algs into doing this job for me?
Any suggestions will be greatly appreciated! I will summarize the
responses to the group in 1 week.
Thank you very much.
regards
jcd
--
Joseph C. Davis, Ph.D. PDF Solutions, Inc.
101 W. Renner Rd, Ste 100 Software and Consulting for
Richardson, TX 75081 Semiconductor Manufacturing
(972) 889-3025 fax: (972) 889-2486
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|