
5.5 Dataplot Links with Other SED Programs
James J. Filliben
Will Guthrie Statistical Engineering Division, CAML
Alan Heckert
Robert Lipman Scientific Computing Environments Division, CAML 1) HPCC/SIMA: The HPCC (HighPerformance Computing & Communication) / SIMA (Systems integration for Manufacturing Applications) project is an omnibus multilaboratory NIST effort to promote and enable stateofthe art integration of computing, systems engineering, and manufacturing. The CAML/SED component is the subproject "Integration of Statistical Optimization Tools in Process/Product Design" This subproject involves the design and construction of an integrated statistical environment for translation and analysis of engineering/manufacturing problems. Dataplot serves as the analytical engine in this environment. A standalone tcl/tk GUI for Dataplot was written (Bob Lipman) as part of this environment. In June of 1995, a demonstration of this environment was given as part of the HPCC/SIMA program review. 2) SEMATECH/NIST Electronic Handbook: There exists a joint CRADA between SEMATECH (the facilitating semiconductor organization) and NIST to produce an electronic handbook, which would be a modernday upgrade to Mary Natrella's NBS Handbook 91 (Experimental Statistics). The working title of the book is "Engineering Statistics Handbook". One important design specification is that the WEB be accessible from Dataplot. Another important design specification is the ability of the semiconductor engineer to not only read about a stat procedure, but also to immediately carry out the analysis on his/her data set. In this regard, Dataplot serves as the role of the analytical engine for carrying out such booksuggested procedures. To achieve this, Dataplot needed to be able to be handbookaccessible with precanned tailormade macros. Such capabilities were imbedded into Dataplot and this integrated system was successfully demonstrated at a SEMATECH meeting. 3) NIST Design of Experiment Short Courses: On a yearly basis, SED teaches a 5day experiment design short course for industry engineers, and (less frequently) for NIST scientists and engineers. Dataplot plays an important role in such a course inasmuch as all of the graphics included in the voluminous course notebook are Dataplotgenerated. Also, a brief dataplot demo is offered as part of the course, and a copy of Dataplot is provided to interested students on a publicdomain gratis basis. 4) NIST Statistics for Scientists and Engineers Short Courses: On a nearannual basis, SED has been offering to NIST scientists and engineers an extensive series of shortcourses on a variety of statistics topics (Susannah Shiller was the previous coordinator; Mark Vangel is the current coordinator). Dataplot is prominently featured in the Exploratory Data Analysis classes of the course. It also contributes less directly as a potential source for some of the data sets used in some of the other classes. 5) SED Software Series: In 1995, SED provided to NIST scientists and engineers a series of 5 talks & workshops on various statistical software packages. The packages included were BMDP, SPSS, Minitab, Statgraphics, and Dataplot. The Dataplot sessions were given April 10 (lecture) and 11 (PC handson workshop). 6) Coatings Consortium: The purpose of the Coatings Consortium (Dupont, Duron, PPG, etc.) is multifold: 1) to establish a mathematical model relating laboratory and field UV degradation field effects; 2) to gain a better understanding of environmental and production degradation factors which contribute to decreased lifetime of coatings; 3) to improve experiment designs at member companies for carrying out companyspecific coating life service life predictions. 4) to improve the design and analysis software tools at the various member companies. Consortium meetings are held 3 times per year here at NIST. As part of the consortium, a handson Dataplot workshop for a UNIX environment was conducted at one meeting; a similar handson Dataplot workshop for PC users was conducted at another meeting. Copies of Dataplot have been downloaded by many member companies. 7) Ceramics Consortium: The purpose of the Ceramics Consortium is to provide improved understanding and methodology in the machining of ceramics. The Dataplot component in this consortium is in the analysis and presentation of results; the techniques for testing underlying assumptions, determining bestfit distributions, evaluating the results of the designed experiments, and creating the presentation graphics for the results were all Dataplotbased. A future component will be the dissemination of the Dataplot software to interested parties among the 21 membercompanies. 8) Standard Reference Datasets Project: The SRD Projectheaded by Eric Lagergrenis an important division and ITL activity whose purpose is to provide benchmark certified computational results for a variety of statistical procedures. Such certified results serve the scientific, engineering, and statistical community in that analysts will thus be able to objectively evaluate the quality of the algorithms and implementations of software packages that they employ. This project has 3 steps: 1. Collect datasets that are representative and/or computationally stressful (pass 1 of this project will have about 60 such data sets); 2. Run these datasets through a battery of extendedprecision tests to ascertain the certified answers; 3. Run the same datasets through a collection of commonlyused software. The Dataplot component to this project cuts across all 3 of the above steps.
Date created: 7/20/2001 