O PnE48O < PO WEEKNAPPLICATION DEVELOPMENT. —| MAY 8, 1989 This ts the last of a three-part series on execu- tive information systems and de- The function of a decision sup- rt system (pss) is to get the right infor- mation to the right people at the right time. h Decision sup- port systems are powerful tools in deci: sion-making. Their role is to help manag: ers gather facts about the internal oper- ation of the organization as well as about competitors and external events. The tools permit managers and analysts to build detailed financial models, ana- lyze these models and use this informa: tion to make decisions. : DSS tools are typically used by finan- cial analysts, business managers, budget managers, product managers and brand " managers. They support sophisticated fi- nancial-analysis functions, including ex- ploratory data analysis, time-series anal- ysis, forecasting, causal models, cross- sectional analysis and advanced analyti- cal tools. DSS products should provide the ana- ' lyst with the facilities necessary to fully v implement a DSS application. The par- ticular type of application and the indi- 3 vidual users determine the mix and depth of the facilities reguired in a sin- gle product. The available DSS products vary widely in terms of facilities. The reguired features of a DSS data / manager are different from those of a production database-management sys- tem. Very often a DSS application uses more than one type of data, and the types of data in an application tend to change over time. š As a result, the data-management fa- cilities of a DSS tool must be very ro- bust. For some applications, a two-di- mensional spreadsheet is adeguate. But to manage both time-series and cross- sectional data simultaneously, many DSS products incorporate a multidi: mensional data manager that allows the user to manage both types of data and view or analyze the data across any dimension or combination of di- mensions. : In addition, these products provide fa- cilities for performing consolidations along any dimension and to analyze or report specific user views. To correctly manipulate the data, the DSS must maintain the associations be- tween data elements, referred to as the. data model, Many of the mainframe DSS products provide automatic navigation tools that ensure that the integrity of the data is maintained, usually through the use of an active dictionary that con- trols all access to the physical database. A eriticism of as Lotus 1-2-3, has been that they do not Use a data dictionary. This can lead to data disintegration: A user could easily operate on monthly information as If it were guarterly data. Micro products - DSS Tools Help Build, Anal - such as Javelin and pcExpress incorpo- APPLIED INTELLIGENCE rate an understanding of business opera- tions and planning and provide data dic- tionarjes that govern access to the data, A DSS tool must be able to manage the diverse types of data and provide data-entry and editing capabilities. To do so, the data must first be loaded into the DSS data manager, as illustrated. Data for a DSS application generally comes from many internal and external sources, some of which:are not machine readable. eb The facilities incorporated. in DSS products to handle these diverse types of data vary greatly. Micro spreadsheet products allow the user to import data How Decision Support Systems Are Stnuctired cility with any DSS product is an inte- grated guery language that is flexible enough to allow the user to specify sim- ple, unstructured reguests. These re- guests tie in directly to the data dictio- nary (if there is one) to ensure that the guery is processed correctly. J Virtually all DSS products provide some degree of report-generation capa- bility, but perhaps more important is the ability to generate graphs. The actu- ' al process of graphical analysis can it- self provide a different insight into re- ported data and can act as a data-re- duction mechanism for guick assimila- tion of information. These advantages have made graphics a key ingredient of The PC Is Executives' Wirnidow Into Company Operatiotis internal Data e Accounting " Marketing % Product Information be External Data % Dow Jones € CompuServe Mainframe Software [2| Data Manager ||, Data Model | ||, li Menu Systemj || Mi PC Software, e Oery and Repo; e Graphics j e Financial Modeling e Statistical Library, e Communications 8 Menu System | e User Interface im Maryelen Zawatski | aged mainframe DSS vendors to imple- A DSS tool must be able to manage diverse types of data and - provide datacentry and editing capabilities. Data for a DSS application comes from manj internal and external sources. from other spreadsheets, from standard databases (such as dBASE III) or text: 'These products generally do not provide data screening, error checking, check-: point/restart or audit facilities or tools for user-defined menus, forms, screen! painters or edit checking. Fi Because of their origins, mainframe DSS products do not have the same limi- tations as the spreadsheet products in the area of data entry and editing. Most. mainframe DSS products provide some sort of standard file definition and input mechanism, and many are able to di: DB2, SGL/DS, IMS and Focus. He Me Perhaps the single most important fa: DSS and EIS applications. With the growing acceptance of the microcomputer as a DSS workstation, graphics are also being used more fre- guently as a dominant user interface, Using a touch-screen or pointing device, the analyst enters responses without us- ing the keyboard. By pointing to an icon or an entry in a list, the user can con: trol the operations of the system. The continued acceptance of graphical inter- face standards, such as the OS/2 Presen- tation Manager, GEM or Topview, will enable more products to provide such interfaces. ii Unfortunately, the field of decision support has commanly been eguated —— DA es ee 'The James Martin Productivity Series, yze Models To Make Decisions with financial modeling or financial analysis. In financial analysis, the user generally manipulates data using a se- ries of mathematical operations and built-in financial functions (such as in- ternal rate of return, present value, payback and growth rate) to better un- derstand the relationships between the data. A financial model allows the user to consistently represent relationships between data elements. While almost all DSS products have fi- nancial analysis and modeling facilities, the implementation of the modeling lan- guages differs greatly. 'The simplest implementation of a fi- nancial model allows the user to devel- op a model in the form of a financial re- port. The financial report becomes the user's view of the financial model. More robust modeling languages allow the user to reference data views not. con- tained in the final report. Procedural logic, such as "if-then-else" or "do while," can be specified as part of a data relationship. Also, more robust DSS products have facilities for multi- level consolidations and the automatic conversion of currency data using differ- ent exchange rates. One component in the DSS architec- ture that varies the most from product to product is the statistical library. Sta- tistical functions available in DSS prod- ucts range from simple mathematical functions (sguare root, absolute value, minimum, maximum and so on) and simple descriptive statistics (mean, me- dian, percentile and so on) to simple correlation and regression, time-series projections, cross-sectional analysis and forecasting routines. A robust DSS product provides the developer with alternative ways to in- teract with an application. Ideally, user interaction should be controlled by the use of a mouse or some other pointing device. Another common method of in- teraction is natural English. Vendors now provide various levels of support for mainframe and micro inte- gration. The continued acceptance of the micro as a user workstation has encour- ment their products in the micro envi- ronment and provide communication ca- pabilities to the mainframe product. Vendors of micro products are also in- cluding facilities for cennecting to main- frames and for data transfers between the mainframe and popular micro products. Next week T'l begin a series of arti- eles on IBM's AS/400 family of midrange Computers as a strategically vital ele- ment of IBM's Systems Application Ar- chitecture for implementing the distrib- uted systems of the '90s. Ni an information service updated guar- terly, is available through High Pro- ductivity Sofivare Ine., of Marble: head, Mass. (800) 242-1240. For infor: mation on seminars, please contact (in the United States ksa! Cemia ) Tech- nology Transfer Institute, 741 10th St., Santa Monica, Calif. 90402 (213) 394 8305. In Europe, contact Savant, 2 New St.,, Carnforth, Lancs, LA5 9BX United Kingdom (0524) 734 505.