Chapter 10 Decision Support and Expert Systems Flashcards
1
Q
Decision Support
A
- An organization’s success depends on the decisions made by employees
- Computer-based systems are beneficial for:
- Large amounts of information
- Intensive processing
- Types of decision support aids
- Decision support systems (DSSs)
- Expert systems (ESs)
- Applications today may combine both types
- Provide single optimal solution or set of solutions
- Decision support modules today may be part of larger enterprise applications
- Also called business analysis tools or business intelligence applications
- Designed to streamline the decision-making process
- -Data warehouses and online processing (OLAP) technologies have enhanced the ability to use data for decision making
2
Q
Computer-based systems are beneficial for
A
- Large amounts of information
- Intensive processing
3
Q
Types of decision support aids
A
- Decision support systems (DSSs)
- Expert systems (ESs)
4
Q
Decision support modules today may be part of larger enterprise applications
A
- Also called business analysis tools or business intelligence applications
- Designed to streamline the decision-making process
5
Q
The Decision Making Process
A
- A decision must be made whenever more than one possible action is available
- It can be difficult to make decisions when many reasonable alternatives are present
- In business, there may be dozens, hundreds, or even millions of different courses of actions available to achieve a desired result
- Decision making is a three-phase process
- Intelligence phase: collect facts, beliefs, and ideas
- Design phase: design the method for considering the collected data, to reduce the alternatives to a manageable number
- Choice phase: select an alternative from the remaining choices
- Businesses collect data internally and externally
- A model is an abstraction of reality, such as:
- Tabletop representations of buildings
- Maps: represent a geographical area
- Mathematical equations representing relationships among variables
- Managers either choose universal models or design their own models
6
Q
Decision making is a three-phase process
A
- Intelligence phase: collect facts, beliefs, and ideas
- Design phase: design the method for considering the collected data, to reduce the alternatives to a manageable number
- Choice phase: select an alternative from the remaining choices
7
Q
Model
A
an abstraction of reality, such as:
- Tabletop representations of buildings
- Maps: represent a geographical area
- Mathematical equations representing relationships among variables
8
Q
Structured problem
A
one in which an optimal solution can be reached through a single set of steps
9
Q
Algorithm
A
a sequence of steps to complete a task
10
Q
Parameters
A
categories of data that are considered in an algorithm
11
Q
Unstructured problem
A
one for which there is no algorithm that leads to an optimal solution
- May not be enough information
- May be a large number of potential factors
- Unstructuredness is closely related to uncertainty
- Examples of unstructured problems
- Weather prediction
- Stock market prediction
12
Q
Semistructured problem
A
one that is neither fully structured nor totally unstructured
- Professionals encounter semistructured problems almost daily in many different industries
- The goal is to choose the one alternative that will bring about the best outcome
13
Q
Decision Support Systems
DSS
A
a computer-based information system designed to help knowledge workers select one of many alternative solutions to a problem
14
Q
DSSs can help corporations by
A
- Increasing market share
- Reducing costs
- Increasing profitability
- Enhancing product quality
15
Q
DSSs consist of three components
A
- Data management module
- Model management module
- Dialog module
16
Q
How do DSS components help users
A
- Enter a request in a convenient manner
- Search vast amounts of data
- Process the data through desired models
- View the results in a desired format
17
Q
Data management module
A
a database or data warehouse that provides data for the intelligence phase
- Accesses the data
- Provides a means to select data by specified criteria
- Many DSSs are intertwined with other organizational systems, including data warehouses, data marts, and ERP systems
18
Q
Model management module
A
turns data into useful information
- May offer a fixed model, a dynamically modified model, or a collection of models
- Dynamically modified model is automatically adjusted based on changing relationships among variables
- Models are used to predict output
Often based on mathematical research - Patterns or models may be unique to a certain industry, such as:
- ATM placement
- Truck route planning
- Airline ticket pricing
- Car rental pricing
- A linear regression model is a general statistical model that is often used
- Gives a best-fit linear relationship between two variables
- A linear relationship can be translated into a program in a DSS
- The actual data points rarely lie directly on the regression line, illustrating the uncertainty
- Regression models are not necessarily always straight lines; they may be curves
- Models often describe relationships between more than two variables
- Some DSSs simulate physical environments
19
Q
Dialog module
A
part of a DSS that allows user interaction with the program
- Prompts the user to select a model and data to process
- Allows the user to change parameters and view the results of the changes
- Displays the results of the analysis in textual, tabular, or graphical format
- Many DSSs are available through the Internet
20
Q
Sensitivity analysis
A
conducted to test the degree to which the total profit grows or shrinks
- If one or more of the factors is increased or decreased
- Results indicate the relative sensitivity of the profit to the changes
- If a small change in a parameter causes a significant change to the outcome:
- Sensitivity of the outcome to the parameter is high
- If the outcome is affected very little by a large change in a parameter:
- Sensitivity of the outcome to the parameter is low
- Also called what-if analysis
- DSS can perform sensitivity analysis on multiple parameters simultaneously
21
Q
Decision Support Systems in Action
A
- DSSs can be used on demand or integrated into a scheme that enforces corporate policy
- DSSs help maintain standard criteria in decision making throughout the organization
- Automated decision production is becoming very popular
- The only labor required is for data entry
22
Q
DSSs are used in many industries
A
- Book sales and food production and retailing: to forecast the number of patrons, the amount of ingredients to purchase, etc.
- Tax planning: tax helper applications such as TurboTax and TaxCut
- Web site planning and adjustment: to analyze shopper behavior, and to design Web sites based on page usage
- Yield management (revenue management): to maximize revenue from airline trips or lodging
- Financial services: to determine loan amounts, and to qualify customers based on credit history
- Benefits selection: to allow employees to make decisions about their benefits
23
Q
Yield management
A
(revenue management): to maximize revenue from airline trips or lodging
24
Q
ES
A
Expert system (ES): emulates the knowledge of a human expert
- Solves problems
- Makes decisions in a relatively narrow domain
- Domain: a specific area of knowledge
- Purpose is to replicate the unstructured and undocumented knowledge of experts, and make that expertise available to novices
- Neural network: a program that emulates how the human brain works
- mimic the way a human brain learns
- Used by more sophisticated ESs
- Constructed with a set of rules, but then it refines itself based on its decision success rate
- Beneficial in detecting fraudulent transactions and claims
- ESs are part of artificial intelligence (AI) research
- AI focuses on methods and technologies that emulate how humans learn and solve problems
- Knowledge base: used by an ES
- Collection of facts and relationships among them
- Uses an inference engine
- Built as a series of IF-THEN rules
- Inference engine: software that combines data input by the user with the data relationships
- Intelligent agent: software that is dormant until it detects a certain event, and then performs a prescribed action
- Case-based reasoning: methodology of solving a new problem established on the solutions of similar problems
- Applied in case-based ESs
- Especially useful in medical decision making
25
Neural network
a program that emulates how the human brain works/learns
- Used by more sophisticated ESs
- Constructed with a set of rules, but then it refines itself based on its decision success rate
- Beneficial in detecting fraudulent transactions and claims
26
Inference engine
software that combines data input by the user with the data
27
Knowledge base
- Collection of facts and relationships among them
- Uses an inference engine
- Built as a series of IF-THEN rules
28
AI
artificial intelligence (AI) research
AI focuses on methods and technologies that emulate how humans learn and solve problems
29
Intelligent agent:
software that is dormant until it detects a certain event, and then performs a prescribed action
30
Case-based reasoning:
methodology of solving a new problem established on the solutions of similar problems
- Applied in case-based ESs
- Especially useful in medical decision making
31
Expert Systems in Action
ESs have been implemented in many industries
- Medical diagnosis
- Help doctors with the diagnosis of symptoms and treatment advice
- Can help enhance the accuracy of Alzheimer’s disease diagnosis
- Medical management
- Help discern patients’ tests and treatments
- Assist with health information management
- Implemented in many industries (cont'd.)
- Credit evaluation
- Used to approve credit card charges
- Used to analyze financial reports submitted with credit applications
- Local loan officers may periodically update the knowledge base adjusting it for current loan policy
- Detection of insider securities trading
- Helps prevent trading of stocks based on private information by analyzing the stock’s history
- Detection of common metals
- Helps nonexperts identify common metals and alloys outside laboratories
- Based on results of simple chemical tests and other information available at the scene
- Diagnosis and prediction of mechanical failure
- Allows vehicle to diagnose its own mechanical problems
- Provides alerts when approaching traffic jam, navigation assistance, Internet service, etc
32
ESs implemented in what industries?
- medical dx
- medical management
- credit eval
- detection of insider securities trading
- detection of common metals
- dx and prediction of mechanical failure
33
GDSS
Group decision support system (GDSS)
- Also called a group intelligence system, collaborative system, or group system
- Facilitates the contribution of ideas, brainstorming, and choosing promising solutions
- Allow participants to define a problem, contribute ideas, then vote on the decision
- Help structure the decision-making process while allowing participants to remain anonymous
- GIS: aid for map-related decisions
- Processes location data
- Web technology helps promote the use of GISs
- Examples: Google Earth, Mapquest, Yahoo Maps
- HTML and XML support the presentation of marked maps
- Used to aid sales and government work
34
GISs are used to help:
- Find shortest paths for deliveries or school bus routes
- City planning for police coverage and health care resources
- Find oil drilling locations
- Locate suitable outdoor recreation sites
- Decide placement of ATMs
35
GIS components
- Database of quantitative and qualitative data
- Database of maps
- Program that displays information on maps
36
Summary
- Decision aids include decision support systems, expert systems, group decision support systems, geographic information systems, etc.
- Three major phases of decision-making process: intelligence, design, and choice
- Types of problems: spectrum from unstructured to structured
- Most DSSs have three components: data management module, model management module, and dialog module
- Sensitivity analysis measures how parameters affect results, and allow “what if” analysis
- Spreadsheets allow users to create DSSs without expertise
- Expert systems are designed to emulate the knowledge of an expert, using artificial intelligence techniques
- Neural network software may be integrated into an expert system to emulate learning
- Expert systems are used in narrow domains where decisions are unstructured
- Geographic information systems are used when decisions involve locations and routes
- Computerized decision aids may overlook important circumstances, leading to inaccuracies or unfairness to individuals
