Computer-Based Applications:Decision Support SystemsVersion 4.0 - 10/18/99
Note to the Student: • Previous lectures have dealt with the theoretical background of decision-making - both at the individual and group levels. • This lecture begins to look at the actual building of application systems for decision support: so-called decision support systems. • Decision Support Systems are abbreviated as DSS.
Quick Reviews • Simon’s Model of Decision Making • Models
Simon’s ModelFlowchart of Decision Process Intelligence Design Choice
Intelligence Phase • Organizational Objectives • Search and SCANNING Procedures • Data Collection • Problem Identification • Problem Classification • Problem Statement
Design Phase • Formulate a Model • Set Criteria for Choice • Search for Alternatives • Predict and Measure Outcomes
Choice Phase • Solution to the Model • Sensitivity Analysis • Selection of best (good) alternative(s) • Plan for implementation (action) • Design of a control system
The role of models in decision-making • A major characteristic of decision-making is the use of models. • A model is a simplified representation or abstraction of reality. • It is usually simplified because reality is too complex to copy. • Basis idea is that analysis is performed on a model rather than on reality itself.
Pounds’ Categories of Models - Expectations against which reality is measured • Historical- expectation based on extrapolation of past experience. • Planning - the plan is the expectation • Inter-organizational - Models of other people in the organization (e.g. superiors, subordinates, other departments, etc.) • Extra-organizational - models where the expectations are derived from competition, customers, professional organizations, etc.
Another classification of models • Iconic Models • Analog Models • Mathematical Models • Mental Models
Iconic and Analog Models • Iconic (scale) models - the least abstract model, is a physical replica of a system, usually based on a different scale from the original. Iconic models can scale in two or three dimensions. • Analog Models - Does not look like the real system, but behaves like it. Usually two-dimensional charts or diagrams. Examples: organizational charts depict structure, authority, and responsibility relationships; maps where different colors represent water or mountains; stock market charts; blueprints of a machine; speedometer; thermometer
Mathematical Models • Mathematical (quantitative) models - the complexity of relationships sometimes can not be represented iconically or analogically, or such representations may be cumbersome or time consuming.A more abstract model is built with mathematics. • Note: recent advances in computer graphics use iconic and analog models to complement mathematical modeling. • Visual simulation combines the three types of models.
Mental Models • People often use a behavioral mental model. • A mental model is an unworded description of how people think about a situation. • The model can use the beliefs, assumptions, relationships, and flows of work as perceived by an individual. • Mental models are a conceptual, internal representation, used to generate descriptions of problem structure, and make future predications of future related variables. • Support for mental models are an important aspect of Executive Information Systems. We will discuss this in depth later.
Decision (Cognitive) Styles • Analytic - planned, sequential approach; learn by analyzing; less emphasis on feedback; formal • Heuristic - learn more by acting than analyzing situations; extensive feedback; intuition, common sense; trial and error. • Autocratic vs. democratic
The Origins of DSS: • The DSS movement grew out of dissatisfaction with two earlier and very successful applications of technology to management: • Operations Research and Management Science (OR/MS) • Management Information Systems (MIS) • By 1970 both technologies were viewed as too limited to: • meet growing demand of managers for more effective decision support • make proper use of the expanding capabilities of information processing technology
Origins of DSS: Problems with OR/MS • The problem with OR/MS was that it was directed to the construction of decision models and to the development of model solution techniques (e.g. in mathematical programming and stochastic processes). • There was insufficient attention paid to the implementation of these models. • No attention paid to the on-going use of models by practicing managers.
Origins of DSS: Problems with MIS • MIS focused too much on support for structured decision processes, rather than semi-structured or unstructured processes. • MIS technology generally used byproduct information from transaction processing systems to provide summary reports for repetitive decision processes.
Origins of DSS: Early Work • The origin of Decision Support Systems (DSS) as a domain of study can be traced back to the late 1960’s at the Sloan School of management at MIT where they studied ill-structured problems. • At the time, general ledger systems, financial planning models, programming languages, databases with query capabilities all came to be referred to as DSS’s. Was DSS just another buzzword? • Contradictory claims and observations abounded on this new concept.
Characteristics ofIll-Structured Problems • The preferences, judgments, and experiences of the decision maker are essential. • The search for a solution implies a mixture of • search for information • formalization, or problem definition and structuring (system modeling) • computation • data manipulation • The sequence of the above operations is not known in advance since: • it can be a function of data • it can be modified, given partial results • it can be a function of user preferences
Characteristics of Ill-Structured Problems - 2 • Criteria for the decision are numerous, in conflict, and highly dependent on the perception of the user (user modeling). • The solution must be achieved in limited time. • The problem evolves rapidly. • Ill-structured problems have many of the same characteristics of the semi-structured or unstructured problems discussed earlier.
Decision Support System Origins: Just Another Buzzword • First there were bookkeeping systems, which made it easy to keep track of things and to generate financial statements. • With the commercial computer came EDP Systems which automated many bookkeeping functions. • Then came Management Information Systems (Management Reporting Systems) which proved so cumbersome and inflexible that management couldn’t use them. • The next panacea of buzzwords came to be known as decision support systems.
Contradictory Claims and Observations about DSS • DSS are interactive systems used directly by managers vs. DSS are typically used by staff. • DSS require special computer terminals and languages vs. DSS can be installed almost anywhere. • DSS projects require careful analysis by highly skilled designers vs. Initial versions of DSS can be built and installed for $10,000. • DSS must be tailored to information needs and personal style of individual managers vs. DSS can be installed to coordinate the efforts of many departments across a corporation.
Origins of DSS: The first DSS • In 1971, under the idea of management decision systems, Michael Scott-Morton implemented a model of the production/distribution network of a major manufacturing company. • The system was the first to do sensitivity (what if) analyses of possible changes in production, distribution, and marketing. • It had two important concepts: • A convenient interactive graphics interface for users. • The collective use of the system by individual managers improved over all organizational effectiveness - the aggregate performance of integrated operations within the firm.
Scott-Morton: Management Decision Systems • The concepts of DSS were first articulated in the early 1970’s by Michael Scott-Morton under the term management decision systems. He defined such systems as “... interactive computer-based systems, which helpdecision makers utilizedata and models to solve unstructured problems...”. (Scott-Morton, 1971).
Keen and Scott Morton: DSS • Keen and Scott-Morton published a seminal book on DSS in 1978. • Their classic definition: • “Decision support systems couple the intellectual resources of individuals with the capabilities of the computer to improve the quality of decisions. It is a computer-based support system for management decision makers who deal with semi-structured problems “ (Keen and Scott-Morton, 1978).
Keen and Scott-Morton: Three Purposes of a DSS • Assist managers in their semi-structured tasks. • Accomplished by providing interactive access to stored data and decision models with a convenient user interface. • Support, rather than replace managerial judgment. • interactive capabilities and convenient user interface allow managers to exert more control over the application of technology • Improve the effectiveness of decision making, rather than efficiency • extend the range and capability of manager decision processes by means of user-friendly interfaces to rapid analyses of decision problems.
DSS: Current Definitions • A DSS is an interactive system that helps people make decisions, use judgment, and work in areas where no one knows exactly how the task should be done in all cases. DSS’s support decision making in semi-structured and unstructured domains, and provide information, models, or tools for manipulating data (Alter, 1995).
DSS: Current Definitions - 2 • A computer program that provides information in a given domain of application by means of analytical decision models and access to databases, in order to support a decision maker in making decisions effectively in complex and ill-structured (non-programmable) tasks (Klein and Methlie, 1995).
The Role of MIS • Management Information Systems: • impact on structured tasks where standard operating procedures, decision rules, and information flows can be readily defined. • Main payoff in improving efficiency by reducing costs, turnaround time, and so on by replacing clerical personnel. • Relevance for manager’s decision making has been mainly indirect, (e.g. providing reports and access to data).
The Role of OR/MS • Operations Research/Management Science: • Impact has been mostly on structured problems (rather than tasks) where the objective data, and constraints can be pre-specified. • The payoff has been in generating better solutions for given types of problems. • Relevance for managers has been the provision of detailed recommendations and new methodologies for handling complex problems.
The Role of DSS in the context of MIS and OR/MS • Decision Support Systems: • Impact is on decisions where there is sufficient structure for computer and analytic aids to be of value but where manager’s judgment is essential. • Payoff is in extending the range and capability of computerized managers’ decision process to help them improve effectiveness. • Relevance is the creation of a supportive tool, under manager’s own control, that does not attempt to automate the decision process, predefine objectives, or impose solutions.
DSS: Working Definition • A DSS is an interactive, flexible, and adaptable computer-based information system that utilizes decision rules, models, and model base coupled with a comprehensive database and the decision maker’s own insights, leading to specific, implementabale decisions in solving problems that would not be amenable to management science optimization models per se. Thus, a DSS supports complex decision making and increases its effectiveness.
Examples of Problem solving with DSS • Firestone Rubber & Tire Company • Houston Minerals Corporation • Portfolio Management • Police-beat allocation in San Jose, California • Mississippi River traffic management • (examples all read/discussed in class)
Idealized Characteristics and Capabilities of a DSS • Provide support in semi-structured and unstructured situations by bringing together human judgment and computerized information. • Support is provided for various management levels ranging from top management to line managers. • Support is provided to individuals as well as groups. • Supports several independent and/or sequential decisions.
Idealized Characteristics and Capabilities of a DSS - 2 • Supports all phases of the decision-making process: Intelligence, Design, Choice • Supports a variety of decision-making processes and styles, e.g. a fit between the DSS and attributes of the decision makers. • DSS must be adaptive over time • DSS must be easy to use. • DSS attempts to improve the effectiveness of the decision rather than efficiency.
Idealized Characteristics and Capabilities of a DSS - 3 • Decision maker has complete control over all steps of the process. It supports, not replaces the decision maker. • DSS leads to learning, which leads to new demands, and the refinement of the system. • DSS should be easy to construct.
Sensitivity Analysis • SensitivityAnalysis - study of the impact that changes in one (or more) parts of a model have on other parts. Generally looks at what impacts changes in input variables have on output variables. • Enables flexibility and adaptation to changing conditions. • Applicability to different situations • better understanding of the model and the problem it supports. • “What-If” Analysis and Goal Seeking
What-If Analysis • Model maker makes predictions and assumptions regarding the input data. • When a model is solved, the future depends on this data. • What If the cost of carrying inventory increases 15%? • What will be the market share if advertising budget increases by 5%?
Goal Seeking • Attempts to find the value of inputs necessary to achieve a desired output level. • Represents a “backwards” solution • If an initial analysis yields profits of $2 million, what sales volume is necessary for a profit of $2.2 million?
DSS Components • Data Management • DSS database • Database Management System • Data Directory • Query facility • Model Management • Model Base • Model base management system • Model Directory • Model execution, integration, and command • Communication (dialogue) subsystem.
DSS: Early Research • Much of the early research on DSS was influenced by the progress in data management (e.g. commercial implementations of hierarchical and network models in the 1970’s, the relational model in the 1980’s). • Much early work attempted to incorporate decision models and user interfaces into data management systems (See Alter’s Classification Schema). • However, later research has seen the emphasis on model management. Data management and dialogue management have many applications outside of DSS.
Model Management • Research on model management began with the suggestion that decision models, like data, are an important organizational resource and that software systems, called model management systems, should be constructed to assist in organizing and utilizing this resource. • The purpose of a model management system is to make the organization and processing of models transparent to the DSS user, just as the purpose of a data management system is to make the organization and processing of stored data transparent to those who wish to maintain it.
Model Management • Model management became viewed as an extension of data management with the result that some information sources were algorithms rather than files. • Current research on relational model management systems includes instances where the output of one model is the inputs of another model.
Model Base Management • Conceptually, the DSS contains a Model Base Management System that manages models and analysis programs in much the same way that a database management system manages data. Besides providing access to a wide variety of models for flexible use, the MBMS should contain: • ability to catalog and maintain a wide variety of models. • the ability to interrelate these models and link them to the database • the ability to integrate model ‘building blocks” • the ability to manage the model base with functions analogous to database management.
Types of Models: Strategic • Strategic Models -use to support top management’s strategic planning responsibilities • tend to be broad in scope with many variables expressed in a compressed form. The models tend to be of a descriptive (simulation) rather than an optimization nature. • Examples: • develop corporate objectives • environmental impact analysis • non-routine capital budgeting
Types of Models: Tactical • Used by middle management in allocating and controlling the organization’s resources. • May be applicable only to one organizational unit or subsystem (e.g. accounting subsystem). • Some are optimization while others are descriptive in nature. • Examples: • labor requirement planning • sales promotion planning • plant layout determination • routine capital budgeting
Types of Models: Operational • Operational Models are used to support day to day working activities of the organization. • Examples: • approving personal loans by a bank • production scheduling • inventory control • maintenance planning and scheduling • quality control
Model Building Blocks • In addition to strategic, tactical, and operational models, the model base could contain model building blocks and subroutines. • Examples: • random number generators • curveline fitting routines • present-value computational routines • regression analysis • All of the above can be used individually for data analysis or combined as components of larger, more complex models.
Communication Dialogue Subsystem • Interface Modes: • Menu Interaction • Command Language • Question and Answer • Form Interaction • Natural Language • Object Manipulation • Interactive Display • Color Graphics • Report Writing
New Directions • Model management, along with data and dialogue management continue to be an important focus of DSS research. However all three are being influenced by developments in artificial intelligence and especially in expert or knowledge-based systems. • Some DSS’s contain knowledge bases and the inferential procedures needed to apply them to a specific decision problem. Examples have been developed for intelligent production scheduling, portfolio management, underwriting, financial statement analysis, diagnosis of equipment failures.