Chapter 10

Decision support systems (DSS) are interactive information systems that assist a decision maker in approaching ill-structured problems by offering analytical models and access to databases. These systems are designed to support the decision-making process, rather than to render a decision. The hallmark of DSS is flexibility. Personal DSSs should be easy to develop: End-user oriented tools are available for the purpose. On the other hand, an organizational DSS, used throughout an enterprise, should be developed in a well-planned, disciplined process. All DSSs should be easy to use.

The Problems that DSSs Help Solve

Problems that people in organizations face differ in terms of how structured the problems are; that is, the extent to which a solution procedure can be stated for them.

The principal domain of DSS is support of decision making for semistructured problems, where parts of the decision process itself often require very significant computer support. DSSs are also employed to assist a decision maker facing an unstructured problem, often injecting the necessary factual grounding through access to data.

Decision making to solve unstructured problems is available but within a narrow domain. An ill-structured problem contains elements of both semistructured and unstructured problems.

A model is a representation of something else, developed for a specific purpose. It is usually an abstraction or a simplification of the phenomenon being modelled. A model represents the relationships among the salient aspects of the phenomenon.

Scenario Generation and Goal Seeking with a DSS

Two principal modes of analysis are available:

1. What-if mode - Developing scenarios for solution with the assistance of information systems (usually decision support systems).

2. Goal-seeking mode - Establishing the input factors necessary to achieve specified goals (usually with a decision support system).

Using DSS in the Decision-Making Process [Figure 10.5]

The general decision-making process consists of four steps. These are:

1. Intelligence - a search of the environment is made to find and define the problem or an opportunity.

2. Design - several alternative solutions are developed

3. Choice - alternatives are compared to one another during the Achoice@ stage.

4. Implementation - solution is implemented and tracked, in order to be improved upon.

Note:

1. Each of the steps may require backing up to a preceding one, in order to redefine the problem or select a better solution.

Features of DSS

DSSs have several features to offer in the general information system environment of an organization. Specifically, DSS can:

1. Support decision making in ill-structured situations when problems do not lend themselves to full computerization.

2. Help to rapidly obtain quantitative results needed to reach a decision.

3. Operate in the ad hoc mode to suit the current needs of the user.

4. Support easy modification of models

5. Foster high-quality decision making

6. Facilitate the implementation of decisions

7. Support group decision making

8. Be user friendly

9. Give managers the opportunity to gain a better understanding of their business

Limitations of using spreadsheets as DSS models include:

1. They are limited in their data-handling capabilities and thus cannot work with large databases

2. They do not allow for construction of more complex models

3. Modifications to spreadsheets are difficulty to keep updated when numerous people use them

The three principal DSS subsystems and their principal capabilities are:

1. Data Management Subsystem

2. Model Management Subsystem

3. The Dialog Management Subsystem

The Data Management Subsystem [Figure 10.7]

Data management subsystem of a DSS supplies data to which the models can be applied. It relies, in general, on a variety of internal and external databases. The power of a DSSs derives from their ability to provide easy access to data.

The database extract procedures used by DSS is generally specified by a specialist, such as a database administrator, rather than by an end user. The specialist needs to pay particular attention to data consistency across multiple decision support systems that extract data from the corporate databases. Data warehouses are used by many leading companies to support organizational DSS. Commercial data warehouses for decision support are emerging.

The Model Management Subsystem [Figure 10.8]

Model management subsystems maintain the libraries of models. A particular advantage of DSS is the decision maker's ability to use a model to explore the influence of various factors on outcomes (a process known as sensitivity). Two forms of such analysis are the what-if analysis and goal-seeking.

The Dialog Management Subsystem [Figure 10.9]

Dialog management model supports the user in applying models to data. The notable feature is support of multiple forms of input and output. By combining various input and output capabilities of a DSS, users can engage in the individually selected dialogs that best support their decision-making styles.

The principal classes of DSS are those that provide:

1. Data access systems

2. Data analysis systems

3. Forecast-oriented data analysis systems

4. Systems based on accounting models

5. Systems based on representational models

6. Systems based on optimization models

7. Systems with suggestion models

Data Access Systems

These systems can provide user-friendly ad hoc access to the database. This capability is equivalent to what is offered by most DBMSs through a query language. However, such systems Aopen-up@ a database.

Data Analysis Systems

These systems help analyze historical and current data, either on demand (ad hoc) or periodically. Data analysis systems are frequently oriented toward the consolidation (aggregation) of data, such as summarizing the performance of a firm's subunits and presenting the summaries in graphs. Only very simple models are employed in data analysis systems.

Forecast-Oriented Data Analysis Systems

These systems generally assist in developing product plans, including market segment forecasts, sales forecasts, and analyses of competitive actions. Their operation is based on access to a variety of internal and external marketing and product databases, including series of historical data. The systems in this category include only the simpler of the variety of marketing models, which show how existing trends in the marketplace will extend in the future if similar conditions prevail.

Systems Based on Representational Models

These models show the dependence between a controllable variable and an outcome. These are frequently simulation models which yield probabilistic results. Examples include representational models and risk analysis models.

Systems Based on Optimization Models

Optimization models are developed by management scientists to determine optimal allocation of resources or best possible schedules.

Systems with Suggestion Models

Systems with suggestion models suggest solutions within narrow domains of knowledge and sometimes combine a DSS with an expert system.

10.5 Building a Decision Support System

DSS technology ranges from the specific DSS developed to solve a class of problems to the tools with which a DSS can be built. Three levels of DSS technology are:

1. Specific DSS

2. DSS Generators

3 DSS Tools

Specific DSS

A specific DSS is the actual system that a manger works with during the decision process. A specific DSS is constructed with the use of DSS generators or a variety of DSS tools. A variety of specific DSS are available in the software marketplace. However, they have to be customized to the actual environment in which they will be used. A DSS generally also undergoes extensive modification as it is used. Therefore, any specific DSS may be expected to evolve as time passes.

DSS Generators

A DSS generator is a software package that provides capabilities for building specific DSSs rapidly and easily. Capabilities of generators vary widely. Their common characteristic is that much of the processing and data accessing functionality needed in a specific DSS is already programmed into the generator and, therefore, building a specific DSS does not require much programming.

DSS Tools

A variety of tools may be employed as building blocks to construct a DSS generator or a specific DSS. These tools include programming languages with good capabilities for accessing arrays of data, simple spreadsheet packages, statistical packages, and DBMSs with a query facility.

How a DSS is Developed

A DSS is a collection of capabilities that support the decision-making process of a certain individual or a relatively small group of people. As the needs of these people change, the DSS should change with them - DSSs are truly built to be changed. DSS can be built by:

1. The quick-hit approach

2. Traditional life-cycle development

3. Iterative development

The Quick-Hit Approach

Characteristics of the quick-hit approach:

1. The quick-hit approach is the way most DSS come into being.

2. Most DSS's are built for the personal use of a decision maker

3. Initiative usually comes form an individual manager, so the DSS is built either by the manager or by the builders who belong to a more or less formal DSS group

4. Generally, a DSS generator is employed, frequently a spreadsheet with templates.

5. Level of investment is low and the payoff can be high

Traditional Life-Cycle Development

Characteristics of the life-cycle development approach:

1. Large software systems are generally built in a disciplined fashion with the use of a life-cycle development methodology

2. This process begins with detailed system planning and analysis, progresses through the design stages followed by coding and testing, and goes on to implementation.

3. Process is lengthy, and there is no partial system to work with before the system is complete.

4. Methodology is suitable for complex systems, in particular those that affect many users and in which informational requirements can be established early through the analysis process.

Iterative Development [Figure 10.11]

Characteristics of the iterative development approach:

1. Develop a prototype of the system - a simple initial version that can be used to experiment with and learn about the desired features of the system.

2. Iterative development of DSS relies on the creation of such a prototype and its progressive refinement.

3. The development of the system is completed jointly with the future user of the system and the DSS builder until the user has a prototype to work with.

4. The iterative, repetitive process of prototype refinement follows until it eventually becomes a DSS.

Group decision support systems (GDSS) are designed to support group communication and decision processes within a group. In the developing information society, more demands are being made for more participation of a number of experts working as a group.

Creativity in Group Decision Making

A number of techniques are being used in work groups to stimulate creative decision making. Some of the techniques include:

Brainstorming - A group decision-making technique aiming at generating ideas.

Synectics - a group decision- making technique where only the best ideas are further considered as the process goes on. Brainstorming aims at fluency in the idea-generating process in order to produce a significant number of ideas.

Nominal Group Technique - addresses the needs of groups in which broad differences of goals and opinions are certain to lead to animosity and defensive argumentation. Therefore, large parts of these sessions are spent by participants working alone, and their ideas are then circulated and evaluated.

Delphi Technique - A method for soliciting the opinions of a group of experts and arriving at a consensus among them. The entire Delphi process is anonymous

Capabilities of GDSS

They help the decision-making group share information, exchange ideas, compare alternative solutions with the use of models and data, vote, and negotiate in order to arrive at a consensus. Such a system consists of a number of software modules acting as tools that support an aspect of this process. It is the objective of a GDSS to enable group members to work simultaneously and anonymously.

Three levels of GDSS capabilities may be distinguished:

1. Level-1 GDSS facilitate communication among group members. They provide the technology necessary to communicate: decision rooms, facilities for remote conferencing, or both.

2. Level-2 GDSS contain the communication capabilities of the Level-1 GDSS and also provide support for the decision-making process.

3. Level-3 GDSS (still in the research stage) will formalize the desired patterns for group interaction, possibly by including expert systems that would suggest rules to be applied during a meeting.

Executive information systems (EIS) provide a variety of internal and external information to top managers in a highly summarized and convenient form. EIS are becoming an important tool of top-level control in many organizations. They help an executive spot a problem, an opportunity, or a trend. Executive information systems have these characteristics:

1. EIS provide immediate and easy access to information reflecting the key success factors the company and of its units.

2. AUser-seductive@ interfaces, presenting information through color graphics or video, allow an EIS user to grasp trends at a glance.

3. EIS provide access to a variety of databases, both internal and external, through a uniform interface.

4. Both current status and projections should be available from EIS.

5. An EIS should allow easy tailoring to the preferences of the particular users or group of users.

6. EIS should offer the capability to Adrill down@ into the data.

Contrasting EIS and DSS: [Figure 10.14]

DSS are primarily used by middle and lower level managers to project the future, EIS's primarily serve the control needs of higher level management.

1. EISs primarily assist top management in uncovering a problem or an opportunity. Analysts and middle managers can subsequently use a DSS to suggest a solution to the problem.

2. At the heart of an EIS lies access to the data. EISs may work on the data extraction principal, as DSSs do, or they may be given access to the actual corporate databases or data warehouses.

3. EISs can reside on personal workstations or servers.

Developing EIS

EIS's should make it easy to track the critical success factors (CSF) of the enterprise, that is, the few vital indicators of the firm's performance. With the use of this methodology, executives may define just the few indicators of corporate performance they need. With the drill-down capability, they can obtain more detailed data behind the indicators.

Strategic business objectives methodology of EIS development takes a company-wide perspective of the strategic business objectives of the firm where the critical businesses are identified and prioritized. Then the information needed to support these processes is defined, to be obtained with the EIS that is being planned. Finally, an EIS is developed to report on the CSFs. This methodology avoids the frequent pitfall of aligning an EIS too closely to a particular sponsor.