Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 241-260(2001)

Knowledge on Internet Environmentg

Michiko Watanabe

(Faculty of Economics, Toyo University)

Katsumasa Suenaga

(Science of Living Department, Kagoshima Immaculate Heart College)

Akinobu Takeuchi

(College of Social Relations, Rikkyo University)

Hiroshi Yadohisa

(Faculty of Science, Kagoshima University)

Kazunori Yamaguchi

(College of Social Relations, Rikkyo University)

Chooichiro Asano

(Faculty of Engineering, Soka University)

The large-scale growth of the Internet as a large-scale communication media has increased information about statistical science and promoted education and efficient dissemination of statistics. This paper outlines the functions of statistical educational Web sites that our project members have been developing as a collaboration research. The first one is the ITLS (Interactive Text for Learning Statistics) site, which is an interactive online textbook for statistics by multimedia tools like JAVA applets. Our newly styled textbook is designed for Web pages and includes a lot of dynamic graphics and images. The second one is the EBSA (Electronic Books for Statistical Analysis) site, which is an electronic book reading system with special searching tools. The third is the DLLSA (Dynamic Link Library for Statistical Analysis) site, which provides various statistical analysis programs that can be called in several application softwares for the Windows system, e.g. Microsoft-Excel. The DLLSA site also provides interface programs over Microsoft Excel. We mention the special searching system only for keywords on statistics.

These educational sites on the Internet have gradually extended both the range of persons who want to study and the chance for studying statistics under every circumstance.

**Key words: Online interactive text, statistical science, electronic book reading system, search engine for statistics, dynamic link library for statistical analysis, educational Web site.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 261-275(2001)

Its Application to Web

Kenichi Satoh and Megu Ohtaki

(Research Institute for Radiation Biology and Medicine,
Hiroshima University)

With the greatly improved development environments on the Windows OS, it is now easy to create various types of applications, such as console applications, applications with graphical user interface and so on. This paper considers how to construct a statistical graph library that can be generally used in these applications. The actual implementation is explained and we show the Web server application in which the library was embedded.

**Key words: Windows OS, Delphi, database, graph library.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 277-292(2001)

Component Analysis to WWW

Masaya Iizuka

(Faculty of Law, Okayama University)

Yuichi Mori

(Faculty of Informatics, Okayama University of Science)

Tomoyuki Tarumi and Yutaka Tanaka

(Faculty of Environmental Science and Technology, Okayama University)

This paper discusses VASPCA/Web, which is a web version of the statistical software VASPCA (VAriable Selection in Principal Component Analysis) for variable selection in principal component analysis. VASPCA/Web is an on-line CGI program written in Ruby and Perl and constructed in our server with the statistical package R as a computation engine. This program is implemented in WWW to provide an environment for performing statistical analysis (variable selection in principal component analysis in this program) and rapidly spreading some new statistical methods so that many users can access them. Users can perform on demand any one of seven selection methods implemented in VASPCA/Web. Some of the methods have been proposed in previous studies and others use new possible selection criteria. The flow of VASPCA/Web is as follows: (1) Entering data (2) Choosing a method (3) Applying ordinary principal component analysis to the data (4) Specifying the number of principal components, a selection criterion and a selection procedure (5) Selecting a subset of variables sequentially (6) Outputting the result (subsets of selected variables and criterion values). This system makes it easy for users in various application fields to perform variable selection in principal component analysis.

**Key words: Statistical program, on-line program, VASPCA/Web.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 293-303(2001)

Yoshiro Yamamoto

(Faculty of Management and Information Science, Tama University)

Tomoyuki Tarumi

(Faculty of Environmental Science and Technology, Okayama University)

Yoshiharu Sato

(Graduate School of Engineering, Hokkaido University)

Even though the statistical services provided on the Web have become general, almost all of the statistical graphics are still static. To provide a dynamic service, we can use Java language. However, there are several plug-ins for Web browsers with dynamic and interactive facilities. It is possible to easily make interactive statistical graphs by using these Internet technologies.

VRML (Virtual Reality Modeling Language) is a modeling language that realizes virtual space with a VRML browser or a VRML Plug-in for a Web browser. By expressing the statistical graphics with VRML facilities, it becomes possible to materialize a method for new graphical application that does not rely on a platform. To make VRML statistical graphics, we have provided prototypes and disclose them to our home page. We have also provided a Web application to make VRML statistical graphics. Finally, we discuss present problems of VRML statistical graphics.

**Key words: VRML, 3D graphics, interactive graphical method.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 305-315(2001)

— An EXCEL Application for Time Series Analysis —

Seisho Sato

(The Institute of Statistical Mathematics)

“E-Decomp” is a statistical software used on EXCEL. Users can easily apply various time series methods, for example, seasonal adjustment, AR fitting, and ARMA fitting. The programs for these methods were originally developed for our Web application, “Web Decomp” (http://www.ism.ac.jp/˜sato/). “E-Decomp” consists of a GUI part and a computational part. The GUI part is written in EXCEL-VBA, and the computational part is written in R and the DLL which is called from R. These two parts are linked using “R-(D)COM Interface”, which was developed by Mr. Thomas Baier (http://www.ci.tuwien.ac.at/R/other-software.html). We use the same programs for the computational part of “Web Decomp” and “E-Decomp”. We can easily develop this new application by using this R-(D)COM Interface. This paper explains how to install and use “E-Decomp” and shows an example of an easy way to develop a new application by reusing existing software.

**Key words: Time series analysis, seasonal adjustment, autoregressive model, Decomp, EXCEL.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 317-331(2001)

— DandD Instance as an Agent on the Internet —

Daisuke Yokouchi

(Department of Mathematics, Keio University)

Ritei Shibata

(Department of Mathematics, Keio University)

We propose InterDatabase, a mechanism to make full use of data scattered over the Internet. InterDatabase also works as an interface to different types of relational databases on the Internet. We describe an implementation of InterDatabase through DandD (Data and Description) instance, in which all necessary information to access data is kept together with enough descriptions of the data itself. A key software is the DandD server, which activates DandD instance as an agent.

**Key words: Data, Internet, DandD,
InterDatabase, XML.**

[ Full text pdf | Back ]

Proceedings of the Institute of Statistical Mathematics Vol.49, No.2, 333-343(2001)

Takeshi Fujiwara

(The Graduate University for Advanced Studies)

Junji Nakano

(The Institute of Statistical Mathematics)

Yoshikazu Yamamoto and Ikunori Kobayashi

(Faculty of Engineering, Tokushima Bunri University)

The computing environment and technologies have been changed and developed dramatically with the popularization of cheap, high-performance personal computers and the Internet. It is indispensable to design and develop statistical systems that can fully utilize them for disseminating statistical techniques and developing computational statistical studies.

We have developed a new statistical system named Jasp (JAva based Statistica Processor) using the Java language that can use many modern technologies easily and efficiently. It has various advanced features, such as a statistical language that has characteristics of both functional based languages and object oriented ones, an interface that integrates a graphical user interface and a character user interface, distributed computing facilities using a network, and powerful extensibilities.

This paper introduces the research background and purpose, and features of Jasp by considering existing statistical systems.

**Key words: Server/client, statistical analysis system, distributed computing, user interface, Java.**

[ Full text pdf | Back ]