I presented a semester's worth of time series topics yesterday - slides
online.

Next week Bahjat talks on the CHANCE article (Syndromic Sureveillance is
it Worth the Effort?)
see Sept. 22 web page box for the scanned copy
http://www.samsi.info/200506/ndhs/workinggroup/ad/
Susie talks on Nov. 3 (multiple testing) and Bahjat again on Nov. 10 -
outlier detection (pdf files also online)

Lisa has agreed to talk on the Shmueli- Fienberg review article.
David is working on getting Cary Priebe to come out and talk about the
Scan Statistics on Enron Graphs
(see online Sept. 22 entries for these)

David and some others of us talked a bit after the session, thinking
that it might be good to just have an informal
discussion of where we are so far and possible avenues of pursuit for
research. Here is my thinking:

Nov. 17 (David - I'd like you to be there and so date could be changed
if needed) General discussion of what
we've heard and where we want to go with it. This might be a short
session.

Nov. 24 this is Thanksgiving, I assume we will not meet.

Dec. 1 Lisa on the Shmueli-Fienberg paper ( Lisa - we can switch with
Nov. 17 if you like)

Dec. 8 and 15 are open and again I assume we will then take a break over
the holidays.

As I mentioned, I would be glad to talk again on unit roots and
stationarity testing. As Myron suggested to me,
I'd say a little more about the difference between maximum likelihood
and least squares and mention a few
reference papers. I'll only do this if you express an interest. I'm not
sure how closely related this is to homeland
security, but a tape of the lecture itself could probably be used at
AbuGraib as a torture device.

Another topic is the P-scan article by Naus and Wallenstein. I've
started reading it and it looks interesting.

One more thing - The stock volume data I mentioned is discussed in the
book SAS for Forecasting Time
Series by Brocklebank and myself, published by SAS publishing here in
Cary and also distributed by Wiley.
Basic ARIMA models and more exotic cointegration and ARCH -GARCH models
are discussed. I meant to
bring a copy and will fo so next time for anyone who is interested.

Let me know what you think about this and especially if you want to hear
some more about time series and
stationarity.

Dave

UPDATE:

OK gang - here's the latest plan:

Oct. 27 Bahjat
Nov. 3 Susie
Nov. 10 Bahjat
Nov. 17 Lisa's brief overview AND discussion of future directions
Nov. 24 Thanksgiving
Dec. 1 me on Unit Roots (probably)
Dec. 8 Curt on the Naus and Wallenstein paper (Curt - we can switch)

Since Lisa's paper is a survey about directions in syndromic
surveillance it will be a good kickoff for discussion.
Curt had a good suggestion that each of us try to look for 2 relevant
papers, maybe for our Nov. 17
discussion, so that a larger pool is available. Let's think about the
holiday break as well.

I am organizing in this way only by default. If anyone wants to suggest
other directions/plans I am open for
suggestions! Have a nice weekend.

Dave