At the beginning of 2011, I made four New Year's resolutions for my blog. As the year draws to a close, it's time to see how I did:
- Resolution: 100 blog posts in 2011: Completed. I blew by this goal by posting 165 articles. I recently compiled a list of my 10 most popular articles.
- Resolution: Provide original content: Completed. Some blogs are primarily aggregators that link to content written by someone else. Not here.
- Resolution: Promote Discussions: I was not so successful here, I think. Although 465 comments were posted during 2011, about 1/3 were responses from me to questions asked by readers. That means readers only posted about 1.8 comments per post, which seems low compared to some blogs that I read. Many articles received no comments.
- Resolution: Learn a new area of statistics: Completed. In 2011 I took a course in time series analysis, learned a little about spatial analysis, and read a boatload of papers on sampling and simulation. Some of these ideas made their way into blog posts.
Is there anything that you'd like to see me do more of in the New Year? I'll post my 2012 resolutions next week.
9 Comments
Rick - keep the DO LOOP rolling. I read and enjoy them all.
Happy holidays.
p.s. when is the SAS glee club VocalMotion going to appear at a SAS Global Forum?
Thanks for the compliments. I am honored.
I suspect VocalMotion's days at SAS Global Forum are over. The last time we appeared was 1997 in Nasheville (SUGI 23). The conference has gotten so big that SAS now books nationally known acts, such as Cirque du Soleil in Vegas last year. However, VocalMotion is going strong and celebrating it's 20th anniversary this year. See vocalmotion.org for a glimpse of what we've been up to recently.
I would really like to see some activity concerning quantile regression. SAS's procedure, proc quantreg, is quite good and very fast, much better than what is in R with respect to performance, but the documentation is a bit sparse. There is no useful discussion about goodness of fit and what to do with various numbers that come along with the quantile regression and so far few user papers for guidance. My area is epidemiology which has done virtually nothing with quantile regression but I am finding it really, really useful, especially with respect to health outcomes and equality issues. What goes on in the tails of distributions is where the action is.
More spatial statistics would be useful. Especially more spatial Bayesian modeling along the lines of what WinBugs does.
This is a useful blog, so are the rest of the SAS blogs. Thanks! Best for 2012.
I don't know if it helps you, but I wrote a paper on PROC QUANTREG for NESUG this year. The PDF is here: http://www.nesug.org/Proceedings/nesug11/sa/sa04.pdf
Keep up the good work! I really enjoy your blog.
For the year to come: a few posts about time series analysis would be great.
What course on time series did you take? It's an area I don't know much about.
Same here. I took a SAS course called "Forecasting Using SAS Software: A Programming Approach," which is taught as a "Live Web" course. I recommend it. It was all about ARMA models and how to use PROC ARIMA.
For other time series courses taught by SAS experts, see http://support.sas.com/training/us/paths/for.html
In your simulation reading does it cover much about time series? I am asking this as in my area of work most time series analysis would be performed to derive a 'goal' for the coming year. This 'goal' is, in essence, a simulation. At least that is what it seem's to me?
Thanks for the link to the Live Web course. I hope to have time to someday check it out!
In 2011 I did not read much about simulation of time series. However, I am sure that much has been written about time series simulation in the economics and financial literature.
SAS/IML software has several built-in functions for simulating ARMA time series. See the documentation of the ARMASIM (univariate) and VARMASIM (multivariate) functions. SAS/ETS software has other tools, and also comes with a GUI application called SAS Simulation Studio, which is used for discrete event simulation and modeling.