NJSUG

30-Mar-2001, Friday, 9:00am, Raritan Valley Community Coll., North Branch, NJ

You can do THAT with SAS Software? Using the socket access method to unite SAS with the Internet
David Ward, President, InterNext, Inc.

The addition of the socket access method to version 6.11 of the SAS System provided the Base, Macro, and SCL languages with a simple way to use the standard TCP/IP network protocol. TCP/IP is the protocol that is used to access virtually all types of network servers, like FTP, Telnet, Web, etc.

In the years since version 6.11 was released, the Internet has grown exponentially and become a vital part of virtually all technology-based businesses. Using the Socket Access Method, the SAS system can interact with the servers that make up the Internet, as well as become a server itself.

This paper will present simple implementations of the following TCP/IP protocols and discuss practical uses for each: HTTP (Hypertext Transfer Protocol), SMTP (Simple Mail Transfer Protocol), POP3 (Post Office Protocol), FTP (File Transfer Protocol), Telnet, and SAS to SAS client/server processing. Uses of these protocols will include building a SAS-based web server, dynamic publishing of SAS data on the Internet without SAS/Intrnet, sending email directly from the SAS system without the need to use a third-party's email software, and reading web pages from the Internet without using the URL access method.

David Ward is president of InterNext, Inc., a software development company based in central New Jersey. He has been using SAS for 4 years and uses it in conjunction with many other programming tools including Perl, Java, Microsoft Visual Studio, and others. He has a B.A. in Math from Rutgers University. He supervised clinical programming at Data Analysis System, Inc. from 1997 to 1999, including teaching and planning SAS training. He the moved into software development in 1999 working for DZS Software Solutions, Inc. David is currently president of InterNext, Inc., a software development company dedicated to SAS software solutions, specializing in Internet applications.

Analyzing the stock market using SAS software
Jim Arvesen, President, Strategic Solutions and Services, Inc.

SAS has been very successful in the areas of clinical development, analyzing sales and marketing data, as well as many financial problems. However, it has not found much use in day-to-day applications in handling the vast amounts of data in the stock market. There are over 1000 stocks that one might want to track on an hourly (or more frequent) basis. In addition to data management, SAS offers the capability of analyzing models associated with these data. Some examples will be presented. Other topics of discussion will include slippage, the difference in the bid/ask spread before and after decimalization, the use of multivariate procedures in SAS, and why univariate procedures are not adequate to the task of finding an edge. Also presented is how SAS can read stock data directly from the Internet, expediting your data acquisition.

Jim Arvesen has an A.B. in Math and Statistics at UC Berkeley, and a Ph.D. in Statistics from Stanford. He taught at Purdue, Columbia, NYU and Fairleigh-Dickinson University, worked at Pfizer for over 20 years, first in clinical development, subsequently in Marketing Research. Dr. Arvesen started Data Analysis Systems, Inc. in 1994 with 4 partners which grew to over 80 staff in 1998, and sold to Quintiles Transnational. He is currently president of Strategic Solutions and Services, Inc. and living on a working farm in Lebanon, New Jersey.

07-Jun-2001, Thursday, 9:00am, Rutgers Labor Education Center, New Brunswick, NJ

Creating and Using Summary Data Sets - Additions In SAS, V8
Ron Cody, UMDNJ, Piscataway, New Jersey

Several SAS procedures produce summary output data sets. PROC MEANS (a.k.a. PROC SUMMARY) and PROC FREQ come to mind. This talk will discuss how to produce and use these summary data sets. Some of the version 8 additions to these procedures will be discussed as well. For example, PROC MEANS now includes a CHARTYPE option and a TYPES statement as well as an AUTONAME output option (which automatically provides names to the various output statistics). You may actually discover the mystery of the underscore variables _TYPE_ and _FREQ_. So, come and learn some new neat stuff!

Dr. Ron Cody is a Professor in the department of Environmental and Community Medicine at the Robert Wood Johnson Medical School, Piscataway, New Jersey. He has been a SAS user for more than 20 years and is the author of "Applied Statistics and the SAS® Programming Language" (fourth edition), published by Prentice Hall. He has also authored or co-authored several books for the SAS Institute as part of their Books by Users (BBU) series. Ron has presented invited papers for numerous local, regional, and national SAS conferences.

SAS Version 8.2 enhancements
Scott Vodicka, SAS, Cary, North Carolina

Scott received a BS in Computer Science from North Carolina State University. Scott was a software developer writing C and C++ applications for 9 years. He joined SAS in 1996 as a Systems Engineer for the Northeast Region focusing on data warehousing, business intelligence, and web Technologies. In 1999 he moved to the North American Marketing department as the Integration Technologies Product Marketing Manager. He moved to his current position as a Technology Strategist in World Wide Marketing in September 2000. Since January he has focused on wireless technologies.

13Sep2001, Thursday, 9:00am, Rutgers Labor Education Center, New Brunswick, NJ

Covariant Method for Analysis of Stability Data
Clayton Rasmussen, New Jersey

This presentation will demonstrate how to compute several regression equations using an Analysis of Covariance (ANCOVA) procedure for various lot and package combinations. It will also demonstrate how to construct estimate statements to compare regression slopes for factors that can affect the rate of degradation.

Additionally, this presentation will address the designing of simulation experiments, which are used to show how random errors affect the distribution of probability values from the estimate statements. By assigning the known regression slopes to the factors that can affect the degradation rate, simulation experiments were generated by adding random errors to the assay values.

Clayton Rasmussen recently retired from Parke-Davis/Pfizer after 31 years in the Product Development department of Warner-Lambert/Parke-Davis. He began using SAS to analyze and present laboratory data when SAS was only a mainframe application, but since then has used SAS on DOS, OS/2, Windows/NT and Unix. In 1989, he began analyzing stability data for regulatory submissions. Clayton used the GLM procedure to compute the regression line and confidence interval and the GPLOT procedure to plot the results. In addition to analyzing laboratory data, he held the leadership role for several laboratory systems, including the Beckman LabManager (Laboratory Information Management System), data acquisition systems (PeakPro and HP LAS), and the Local Area Network in Product Development.

An Application of PROC NLP to Survey Sample Weighting
Talbot Katz, New Jersey

For surveys and other studies, we may need to analyze or model a random sample of a large population. In order to extrapolate the results in the sample back to the original population, subgroups of the sample must be weighted in proportion to their size (or other importance measure) within the original population. (We assume no sample bias or stratification.)

Typically the weight for each individual in sample subgroup i is equal to (p{i}/P)*(S/s{i}), where P is the size of the original population, S is the size of the sample, p{i} is the size of subgroup i in the original population, and s{i} is the size of subgroup i in the sample; the sum of these numbers over the whole sample is equal to S (the weights of the subgroups in the sample are redistributed in proportion to the weights of those subgroups within the population). This weighting formula does not apply when subgroups of the sample are empty.

Usually empty subgroups are not important by themselves, and sometimes such subgroups can be folded in with non-empty subgroups. But there are cases when this is not practical. For example, suppose we are studying a population of IT professionals that has both employees and consultants, each of whom can be either a SAS user or unenlightened (i.e., a non-SAS user). If our sample of consultants who are not SAS users is empty, and if we wish to study differences between employees and consultants and differences between SAS-users and non-SAS-users, we cannot combine the weight of the non-SAS-using consultants into any of the remaining subgroups. A quadratic minimization algorithm is presented, via PROC NLP, to weight all the non-empty sample subgroups in such a way as to retain the most important properties of the original population. A brief introduction is given to PROC NLP, including discussion of some options which allow flexible treatment of this problem.

Talbot Katz is a SAS consultant who has modeled for the telecommunications, credit, and pharmaceutical industries (they have remarkably similar wardrobes!). He has been a member of NJSUG for a random sample of the past eight years.

06Dec2001, Thursday, 9:00am, Rutgers Labor Education Center, New Brunswick, NJ

Intro to the Macro Map
Russ Lavery, Saad Anbari, Musa Nsereko, Philadelphia, PA

Macros recall text and conditionally execute code
The three SAS compiles and the three SAS executes
Review of tokenizaton and regular SAS job
Assigning/resolving a macro variable (what causes errors)
Macro Compiling/Execution of a Macro Program (what causes errors)
Evaluating && and &&& and &&&&

The Macro Reference Environment and Methods of Influencing Macro Execution
Russ Lavery, Saad Anbari, Musa Nsereko, Philadelphia, PA

The macro referencing environment
Global and local macro variables
Logic for update vs. creation of macro variables
%include
call execute
Macro looping (%do i = 1 to &num)

Russ Lavery is an independent contractor who has used SAS for 15 years. He specializes in using SAS and statistics to solve business problems.

Saad Anbari is a SAS programmer for Sanofi-synthelabo and has used SAS for 5 years. Saad has experience in using SAS in the health care industry. SAS Certified Professional.

Musa Nsereko is is a SAS programmer for Sanofi-synthelabo and has used SAS for 7 years. Musa has experience in using SAS in the health care industry. SAS Certified Professional.