This section presents several recommendations for future work with TABIS. Some of the recommendations could be implemented using the existing TABIS infrastructure, while others entail additional development work. Hopefully these ideas will challenge and motivate others to continue with the TABIS research work.
The first step should be "advertising" that TABIS is available to students and faculty at NCSU. The advertising could be in the form of special seminars, person-to-person recruiting, a promotional video or pamphlet, or assigning the use of TABIS in class projects.
After the potential users are aware that TABIS exists, they will need basic training - a short brochure, or on-line help file would probably suffice. Also, a consultant that is familiar with TABIS should be available to help handle problems which might arise.
One of the goals of the original NTC project which launched TABIS was to help enhance the competitiveness of U.S. apparel manufacturing firms. As TABIS reaches maturity, reports and articles could be generated from TABIS which would be very useful to apparel manufacturing firms. A plan should be developed to ensure that such information is generated, and passed on to interested firms using whatever means might be most effective.
As textile and apparel manufacturers become aware of TABIS, through the reports and articles, or by word of mouth, they will probably want direct access to the data so they can perform their own analyses. Such users could access TABIS over the Internet, or through dial-in connections, but provisions will need to be made to set up accounts for such "outsiders" (people who are not NCSU students or faculty) so they can use the NCSU computers. Alternatively, methods could be developed which allow them to access the data without an actual account - special techniques using e-mail, and World Wide Web servers (via SAS/Intrnet), are currently under development at SAS Institute which might prove useful in this endeavor.
Although the ASCII interface is very versatile and simple to use, benefits could be gained by developing a graphical user interface (GUI). New and emerging SAS products like SAS/AF (Application Frame) and SAS/GIS (Geographical Information System) could be used to provide GUI interfaces, with features like buttons, scroll bars, selection lists, and hyper-graphics.
The selection lists, for example, could allow users to dynamically scroll through actual data and click on values of interest, instead of using static values hard-coded into the ASCII interface. This would eliminate the need to re-program the TABIS menus each time new data arrives which contains additional selections -- the selection lists could be generated dynamically from the data.
Emerging products, like SAS/GIS, allow users to interact with graphical plots and maps directly. For example, while looking at the map of the U.S. textile employment by county, the user could click on a particular state to have that state displayed full-screen, and then click on a particular county within that state to see the actual textile employment for that county. These hyper-graphic (graphics with hyperlink) GIS capabilities are not available in SAS 6.09, which is currently installed at NCSU, but are available in version 6.11, which has recently been released and should be available on the NCSU computers soon.
The TABIS data selection capabilities could also be integrated with SAS's new and emerging GUI data analysis products such as SAS/Insight, SAS/Spectraview, and JMP. TABIS could be used for the data selection phase, and then the selected data could be easily fed into the new user-friendly GUI interfaces which are specifically designed for data exploration and analysis.
Also important will be loading "fresh" data into TABIS as it becomes available in the future. Currently, this is a manual job, in which a copy of the data must be obtained, loaded into our local computers, and merged into the TABIS data warehouse (see Section 4).
In the future, if the data becomes available on the Internet, TABIS should be enhanced to automatically receive fresh data directly from the source(s). This way the manual process could be eliminated, and TABIS would always have the freshest possible data.
In addition to keeping the existing data fresh, there will be a demand for new data and reports in TABIS. Although both tasks could be accomplished with user-written extensions, there will be cases when new data and reports that would be of interest to many people should be added to TABIS, instead of expecting each user to write the same extensions.
Another area for future work is the development of more sophisticated forecasting models. The framework for developing such models is already in place with the data selection capabilities of TABIS, and the traditional forecasting procedures which are built into the SAS language. Research could be performed, applying the built-in forecasting procedures to the data, and developing very useful models - this would be a good topic for graduate research work at the Master's level.
In addition to the traditional forecasting procedures, SAS Institute has recently released new tools which can utilize Neural Nets (see Section 2.3.5) for forecasting. When NCSU obtains the new release of SAS, these tools could provide some interesting results.
In addition to applying the forecasting procedures which are built into SAS, TABIS allows enough flexibility to develop new forecasting techniques. Developing such new techniques could provide the basis for graduate work at the Ph.D. level.