I am Roy Pardee, Data Wrangler, Analyst and Application Developer from the greater Seattle area. I've been doing various bits of computing in the service of research (mostly Health Services research) since 1997 or so.
In my little professional circle I am probably best known for my work designing and implementing the HCSRN's Virtual Data Warehouse.
This page is intended to be something of a résumé and calling-card for anyone interested in my professional life.
At my work we frequently do multi-site research and then transfer data from one site to another. That poses the risk that we will send more data than we intend, with the dire HIPAA implications of having "disclosed" PHI. This macro runs through a set of datasets looking for tell-tale signs that they include PHI. This macro and another (which I did not write) were written up in Medical Informatics & Decision Making.
This script identifies orders for PSA tests for men who probably don't need them, and sends an e-mailed list to the relevant Service Head. This is part of the Choosing Wisely initative work that Group Health is doing.
A really interesting study of an intervention in cancer treatment needed to recruit patients immediately after they learned of their cancers. At the time the best method of identifying possible recruits was to read pretty much every pathology report that came through our system, which we didn't have budget for. Having recently read Programming Collective Intelligence I was able to put together (and train!) a Fisher text classifier that filtered the stream of path reports down to something we could manage.
I took a coursera course on this that I absolutely loved (though it was way more time commitment than expected) and would love to get more into this sort of work—both the "big data" aspects and the high-level analytics.
Best. Editor. Evar. One of my proudest accomplishments is the sublime package I put together for SAS programmers.
Pretty much my bread-and-butter at the moment.
Gorgeous, lovely, wonderful programming language.
Which is in fact two things--powershell (which, once I got past my resentment of microsoft at making me learn another language is actually quite cool) and git.
(served from github.io)