SAS® Statistics and
Operations Research News 		SAS
June 2011 support.sas.com/statistics/  |  subscribe  |  unsubscribe
Baseball and Statistics

Dear Readers -

Well, I made up for missing spring training baseball with a 14-inning marathon at Fenway Park in Boston on June 4. That jogged my memory, and sure enough, I found  my handwritten scorecard for a 16-inning game on the same day 45 years earlier! (The Sox defeated the Yankees 6-3)

Sports statistics sent me down the path of mathematics and statistics – what grabbed your interest? This is timely as SAS hosts students from the Summer Institute for Training in Biostatistics (SIBS) at North Carolina State University this month for an afternoon of demos, talks, a career panel and a mixer. This program exposes undergraduates with quantitative interests to possible careers in statistics, and I'm a big fan. We have a blast with the kids every summer.

It's nice to be readying another release of the analytical products just one year after the 9.22 releases went out the door. We are on target to ship the 9.3 release this summer, which means new 9.3 releases for SAS/STAT®, SAS/OR®, SAS/QC®, SAS/ETS® and SAS/IML®, as well as new releases for our "cousins" in the data mining arena. So, you might want to pay attention to the support.sas.com site for availability and encourage your IT personnel to load the latest release of SAS® software as soon as they can.

Well, I did see an Elvis or two (and one Spiderman) at SAS® Global Forum in Las Vegas, and I followed that up with a talk for the Harvard Medical School In-House User Group before I headed to the Iowa Users Group  One-Day Conference in Des Moines. The Iowa organizers did a great job of putting together two tracks, and the attendees received some great information. Next on my conference list is the Joint Statistical Meetings in Miami Beach, FL, from July 31 to August 4. Please come by the SAS booth if you attend that conference – we'll be demoing lots of features from our recent releases (and giving away an iPad to boot!).

Finally, I'd like to announce that we'll be shifting to a bimonthly e-newsletter in the near future, and we are gearing up to deliver even more content. Please let me know if you have specific ideas for inclusion.

Stay tuned,
Maura

R&D Research Director, Statistical Applications
SAS News
Creating Funnel Plots with SAS/IML® Software »
Master SAS/IML blogger Rick Wicklin discusses how to create funnel plots, which allow you to display estimates of quantities against some measure of their precision.  If you aren’t aware of this blog, you might browse Wicklin’s other postings as well – it’s great stuff! And if you’d like to start learning about SAS/IML, Wicklin has suggestions for how to get started.
SAS/STAT® 9.3 Teaser »
For a preview of the many analytical product updates with SAS/STAT 9.3, take a look at what’s coming in SAS/STAT 9.3. Building on the many enhancements in SAS/STAT 9.22, SAS/STAT 9.3 delivers exciting new functionality. The experimental  FMM procedure fits models to data where the distribution of the response is a finite mixture of univariate distributions.  The SURVEYPHREG procedure becomes production and now  handles time-dependent covariates. The PHREG procedure supports frailty models, and the MI procedure offers additional flexibility by including a fully conditional specification method. PROC MCMC now provides a RANDOM statement to facilitate fitting Bayesian models with random effects. SAS/STAT 9.3 contains numerous other enhancements as well. 
Linear Optimization in SAS/OR® Software: Migrating to the OPTMODEL Procedure »

PROC OPTMODEL, the flagship SAS/OR optimization procedure, is intended to supersede the INTPOINT, LP and NETFLOW procedures for linear optimization. PROC OPTMODEL's rich and flexible syntax enables natural and compact algebraic formulations of optimization problems. And the linear programming, mixed-integer linear programming and network solvers available in PROC OPTMODEL are much faster than those in the INTPOINT, LP and NETFLOW procedures.

Beyond the easier modeling and access to improved algorithms, PROC OPTMODEL also provides programming capabilities that enable you to develop customized solution methods. This paper uses several examples to illustrate how to migrate to PROC OPTMODEL.
Small Area Estimation for Survey Data Analysis Using SAS Software »
Small area estimation is important in survey analysis when domain (subpopulation) sample sizes are too small to provide adequate precision for direct domain estimators. Popular techniques for small area estimation use implicit or explicit statistical models to indirectly estimate the small area parameters of interest. Indirect estimation requires you to go beyond the survey data analysis methods that are available in the SAS/STAT survey procedures. This paper describes the use of the MIXED, IML and MCMC procedures to fit unit-level and area-level models and to obtain small area predictions and the mean squared error of predictions. Hierarchical Bayes models are also discussed as extensions to the basic models.
CONTRAST and ESTIMATE Statements Made Easy: The LSMESTIMATE Statement »
SAS/STAT software provides a wealth of facilities for postfitting inference for its modeling procedures. The CONTRAST and ESTIMATE statements provide the means for constructing customized hypothesis tests, and the newer LSMESTIMATE statement enables you to specify hypotheses in terms of the population quantities of direct interest. This paper provides advice about the use of these statements and their newer features, including the new nonpositional syntax of the LSMESTIMATE statements. Examples come from actual user questions to SAS Technical Support.
Talks and Tutorials
Joint Statistical Meetings 2011 »

The statistical R&D division will present several courses at the Joint Statistical Meetings in August.

Fang Chen presents a one-day course on Practical Bayesian Computation on Monday, August 1.

Warren Kuhfeld presents a two-hour tutorial on The Graph Template Language in SAS and the Statistical Graphics Procedures on Wednesday, August. 3.

Rick Wicklin presents a two-hour tutorial on Data Simulation for Evaluating Statistical Methods in SAS on Wednesday, August 3.
Tech Support Points Out
Tests for Comparing Nested and Nonnested Models »
Compare two nested or nonnested models fit by maximum likelihood. Nested models are compared using the likelihood ratio test. Nonnested models are compared using tests by Vuong or Clarke testing the hypothesis that both models are equally distant from the true model.
Support
Quick Links
FASTats: Frequently Asked-For Statistics »
Newsletter Archives »
SAS Discussion Forums »
Getting Started Resources for SAS/STAT 9.2 »
New Product Pages on the Support Site »
SAS/STAT® Procedures A-Z »
SAS/STAT® 9.22 Resources »
 
Don't miss important updates from SAS! Please add sas.com as a domain
in your safe sender list.
Contact Us
Have a question, comment or suggestion? Want to see something covered in the next SAS Statistics and Operations News? Simply e-mail the editor.
 
Tech Support | Privacy Statement


SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks
of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and
product names are trademarks of their respective companies.

Copyright © SAS Institute Inc. All rights reserved.