CLASS-L Archives

January 2009


Options: Use Monospaced Font
Show Text Part by Default
Condense Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"Classification, clustering, and phylogeny estimation" <[log in to unmask]>
Mon, 26 Jan 2009 20:12:13 -0500
"Classification, clustering, and phylogeny estimation" <[log in to unmask]>
text/plain; charset=ISO-8859-1; format=flowed
Steve Peck <[log in to unmask]>
text/plain (62 lines)
It would be interesting know what you came up with 'clustering across time'
but I think we'll more easily understand the relations b/t Procedure 
Codes and Treatment Behavior (if I got this right)
if we cluster first within moments of time (where all the substantive 
dynamics come closely together)
 and then link these profiles (or configurations) across time (eg 
crosstabs or loglinear models).
In this way, we can observe changes over time in the "sample-level" 
range of Procedure Codes X Treatment Behaviors
(eg, maybe Code x Behavior configurations will become stronger over time 
or different configurations will emerge over time)
  and we can also observe individual-level changes over time, as each 
person moves into and out specific CxB configurations
  (eg, some personal pathways may be more probable than others, and 
there may be other variables that can predict these pathways or pathways 
that deviate from the most probably pathway)
...just some thoughts...


Shannon, William wrote:
> I have a new project that is intriguing and offers data I have never tried to cluster.
> The question is whether there are hiv/aids patients treated differently according to their medicaid procedure codes (what was done to them). Procedures occur over time so  changing the clustering over time would seem reasonable. Hopefully we will see clusters of patients form based on medical interventions given, and stabilize as we move forward.
> I have never done anything like this and wanted to know if anyone could point me in the right direction towards previously done work along this line.
> Thank you
> Bill Shannon, PhD
> Associate Prof. of Biostatistics in Medicine
> Washington University School of Medicine
> Director, Biostatistical Consulting Center
> 314-454-8356
> ----------------------------------------------
> CLASS-L list.
> Instructions:

Stephen C. Peck
Research Investigator
Achievement Research Lab
Research Center for Group Dynamics
Institute for Social Research
University of Michigan
426 Thompson Street, # 5136
Ann Arbor, MI  48106-1248
(734) 647-3683; fax (734) 936-7370
[log in to unmask] 

CLASS-L list.