 |
|
|
|
California, Here We Come
Ready or not, a whole bunch of SAS® users and SAS staff will soon walk the streets of San Francisco, chancing vertigo on the Golden Gate Bridge, looking sideways at some alternative scenes, visiting Cannery Row, strolling around Pacific Heights, and turning the conversation from SAS® Visual Analytics to the evening’s restaurant hunt. Beyond a shadow of a doubt, the graduate of a pre-conference statistics course and a fearless SAS macro veteran will hurry from their lunch in bustling Chinatown to the room for afternoon sessions with more urgency than an escape from Alcatraz. In the final analysis, if you are attending SAS® Global Forum this year, you'll have the time of your life!
If you are already registered or still on the fence (early-bird registration ends March 25), you may be interested in these conference paper highlights and demo highlights. And for you Operations Research folks, SAS Global Forum includes a session dedicated to your applications on Wednesday morning.
Speaking of Operations Research, the INFORMS Conference on Business Analytics and Operations Research is slated for April 7–9 in San Antonio, Texas. It includes a number of SAS speakers, including Ayşegül Peker, presenting "Win-Win in Inventory Management Using Advanced Analytic Techniques," and Mike Gilliland, presenting "Process Control Approaches in Business Forecasting," as well as pre-conference workshops and software demonstrations. If you attend, please stop by the SAS booth on the exhibition floor.
Here's hoping we see some of you this spring!
Maura
Senior R&D Director, Statistical Applications
|
|
New Features in SAS/STAT® 12.1 »
If you finally have the 12.1 release of SAS/STAT software on your PC and are ready for action, this paper will get you up to speed on its new capabilities, such as model selection for quantile regression, quantile regression for censored data, and multivariate adaptive regression splines. Learn about using the FMM procedure for finite mixture models, managing missing data with the MCMC procedure, and much more. Better catch up now—rumor has it that development is working fast on the next major release of SAS/STAT software!
|
Thoughts on Statistical Education »
The Year of Statistics, an effort to educate the public about the value of statistics, is getting a lot of attention in 2013. But an equally important issue is how well we are training statisticians themselves for today's data challenges. Many of you take or teach graduate courses, and I thought you might be interested in ASA President Marie Davidian's current column in Amstat News, "Doctoral Training in Statistics and Biostatistics: Where Are We Headed?"
|
|
|
Using the HAZARDRATIO Statement in the PHREG Procedure »
If you are learning to use the PHREG procedure, you might want to find out more about using the HAZARDRATIO statement to estimate hazard ratios. This paper provides an overview, and also discusses the new CONTRAST statement and CLASS statement. The emphasis is on illustrative examples of comparisons for main effects and interaction models via the new statements.
|
|
|
Fitting Frailty Models with the PHREG Procedure »
Failure times can be correlated—for example, if they come from the same cluster—and this correlation must be accounted for in any analysis. The frailty model is one way to handle this correlation, and it does so by using a random component for the hazard function. This video tells you how to fit this model by using the PHREG procedure.
|
|
|
Logistic Regression Using SAS: Theory and Application, 2nd Edition »
In Logistic Regression Using SAS: Theory and Application, 2nd edition, Paul Allison illustrates the modeling of each type of categorical response—binomial, ordinal, and multinomial. Along the way, he shows you the new features and capabilities of PROC LOGISTIC, such as exact estimation, ROC analysis, odds ratio estimation for models with interactions, new graphing capabilities, and more. Allison also addresses how to make use of features from other procedures, such as PROC MDC for discrete choice modeling and PROC QLIM for estimating marginal effects. In addition to logistic regression, he covers the related problem of modeling a count response by using Poisson and negative binomial regression. Common modeling problems, such as separation, multicollinearity, and overdispersion, are discussed and possible solutions are presented. Allison focuses on the practical application of these methods, with attention to the interpretation of results. He offers plenty of examples and writes in a conversational, easy-to-understand style. (David Schlotzhauer, Tech Support Statistician)
|
|
|
SAS Global Forum 2013 »
The following tutorials will be presented on April 28 at the 2013 SAS Global Forum in San Francisco.
- Modeling Categorical Response Data (Maura Stokes)
- Model Selection with SAS/STAT Software (Funda Güneş)
- Introduction to the MCMC Procedure in SAS/STAT Software (Fang Chen)
- Creating Statistical Graphics in SAS (Warren Kuhfeld)
|
|
Comparing groups with respect to the dose (or predictor value) that yields a specified response probability »
When you are modeling the probability of an event, p, the INVERSECL option in PROC PROBIT provides estimates of a continuous predictor (often a dose variable) that yield a range of response probabilities. Suppose you have several groups of subjects and you fit a model to each group. You might want to compare groups with respect to the dose that yields, say, a 50% response probability (commonly called the ED50 or LD50). This note discusses and illustrates how tests and confidence intervals of differences of such dose values can be obtained.
|
|
|
