Decision Support and Expert Systems

1
Decision Support and Expert Systems

• Chapter 10 Oz (5th ed)

2
Decision Making in Business

• Organizations success depends on quality of
  managers decisions
• When decisions involve large amounts of data and
  complex processing, a DSS is a valuable tool
• When decision making involves many uncertainties
  and/or lots of alternatives a DSS is needed

3
Simons Stages of Decision Making

• Intelligence - the individual collects
  information to identify problems occurring
• Design - the individual conceives of possible
  alternative solutions to the problem
• Choice - the individual selects among the various
  solution alternatives
• Implementation - the individual puts the decision
  into effect and reports on the progress of the
  solution

4
The Decision Making Process Is Not Linear
Intelligence
Is there a problem?
Design
What are the alternatives?
Choice
Which should you choose?
Implementation
Is the choice working?
5
Typical Activities in the Stages of
Decision-Making
6
Structured and Unstructured Problems
7
Types of Decisions

• Structured decisions are repetitive, routine, and
  involve a definite procedure for handling
• Unstructured decisions are nonroutine decisions
  in which the decision maker must provide
  judgment, evaluation, and insights there is no
  agreed-upon procedure for making the decision
• Semistructured decisions are ones where only part
  of the problem has a clear-cut answer provided by
  an acceptable procedure

8
Example of a Structured and Semistructured Problem

• Structured problem How much will I earn after
  two years if I invest 100,000 in municipal bonds
  that pay 4 percent per annum tax free?
• Semistructured problem If I invest 100,000 in
  stock XYZ and sell the stock in two years, how
  much money will I make?
• How are these problems different?

9
Examples of Structured and Semistructured Problems
10
A Definition of a DSS

• A DSS is a computer system that combines data,
  sophisticated analytical models, and
  user-friendly software to support semi-structured
  and unstructured decision making.
• Examples
• Some DSS are very sophisticated and built by
  experts to improve firm performance (e.g., in
  trucking, models are developed to minimize total
  mileage and maximize truck loads while
  maintaining satisfactory delivery times)
• Other DSS are built for individual managers to
  assist in daily decision making tasks (e.g., Mrs.
  Fields Cookies provides managers with a DSS that
  assists in decisions related to which cookies to
  make, how many to make based on prior sales)

11
Characteristics of a DSS

• DSS offer users flexibility, adaptability, and a
  quick response (usually interactive)
• DSS allow users to initiate and control the input
  and output
• DSS operate with little or no assistance from
  professional programmers (e.g., personal and
  CDSS)
• DSS provide support for decisions and problems
  whose solutions cannot be specified in advance
• DSS use sophisticated analysis and modeling tools

12
How a DSS Can Improve Managerial Decision Making

• In the intelligence phase
• produce more timely data
• increase the use of quantitative data
• make more data available
• In the design phase
• generate more alternatives
• generate more accurate alternatives
• In the choice phase
• make faster decisions
• sometimes rank the choices

13
Components of a DSS

• Data management module
• consists of current or historical data and
  possibly some external data data warehouse
• Capabilities to manipulate the data
• Model management module
• collection of mathematical models (statistical
  models, linear programming, forecasting,
  sensitivity analysis models) available for the
  user
• May also include OLAP or datamining tools
• Dialog management module
• Interface between user and other modules
• Displays the analysis of results

14
Typical Features of Commercial DSS

• Easy to use graphical user interface (Windows
  like)
• Drill down capacity
• On-demand financial analysis (ratios or
  indicators)
• Access to external (outside the firm) databases
• Statistical analysis tools
• Tools to navigate in a database or data warehouse
• Ability to perform ad hoc queries or what-if
  analysis
• Ability to solve diverse business problems

15
Web Sites With DSS Products

• Examples of DSS software
• http//cdss.state.co.us/
• http//www.vanguardsw.com
• DecisionScript enables you to build Web-based
  DSS or CDSS
• DecisionPro can be downloaded for free 15-day
  trial
• Examples of EIS
• http//visualmining.com
• Look at dashboards and specific examples
• http//www.meditech.com/productbriefs/pages/Produc
  tBriefsCSESS.htm
• Drill down, customized views, individualized
  reporting

16
Types of DSS (one way of classifying)

• Model-driven DSS are primarily standalone systems
  that use some type of quantitative model to
  perform what-if or other types of analysis.
• Data-driven DSS are systems that analyze large
  pools of data that are (usually) stored in a data
  warehouse these systems permit managers to
  extract useful relationships that exist in the
  data (datamining).
• Web-based DSS exist for both types (customer
  decision support systems Vanguard, Rainbird
  irrigation systems), Home Depot (deck projects)

17
Other Types of DSS

• Geographic Information Systems (GIS)
• Customer DSS (CDSS)

18
Geographic Information Systems (GIS)

• A GIS is a DSS for decision aid for map-related
  decisions
• Typical GIS contains
• Database of quantitative and qualitative data
• Database of maps
• Program to display information on maps
• Applications of GIS
• Identifying the best site to locate a branch
  office
• Targeting pockets of potential customers
• Repositioning promotions and advertising
• Tracking the spread of disease or crime

19
Web Sites for GIS

• HUD environmental maps http//198.102.62.140/emaps
  /SearchFrame.asp
• A major GIS vendor
• http//maps.esri.com
• An example of how a state can use a GIS
• http//www.state.ma.us/mgis/muniapps.htm
• An example of how Charlotte, NC uses a GIS
• http//meckcama.co.mecklenburg.nc.us/relookup/ /
• http//gischamber.co.mecklenburg.nc.us/website/cha
  mber/Default.aspx
• An example from San Francisco, CA
• http//www.sfgov.org/site/gis_index.asp
• Use APN 8719003 in SFprospector demographic and
  business report for Mission Bay South location
• An example from York County, SC
• http//www.yorkcountygov.com

20
Customer DSS (CDSS)

• A CDSS is a special type of DSS designed to
  support the buying decisions of an existing or
  potential customer. Most current CDSS are Web
  based
• Support exists for both the individual consumer
  (B2C) and businesses (B2B)
• Complexity of products and sophistication of the
  buyer often determine the nature of a CDSS

21
Examples of CDSS

• BusinessWeek (www.businessweek.com/bschools)
• PriceGrabber ( www.pricegrabber.com)
• BankRate (http//www.bankrate.com/)
• HomeDepot (http//www.homedepot.com)
• Design tools and calculators for various projects
  
• Automobile tires (http//www.tirerack.com)
• Assistance for choosing tires for an automobile
• Vanguard (Vanguard - Investing for retirement -
  Determine how much to saveearlyearly)
• Retirement planning tool

22
Manager and Consumer Decision Making
Managers Decision-Making Process Customers Purchasing Decision Process Description
Intelligence Need recognition Customers must recognize they have a need
Intelligence Information search Customers must search for a product or service that satisfies this need
Design Evaluation Customers use decision heuristics to evaluate and compare goods and services
Choice Purchase The customer must place an order, pay for it, and arrange for collection delivery, or maybe installation
Implementation After purchase evaluation Customers evaluate purchases with a view to future decision-making. Customer support and maintenance may be important
23
The Electronic Spreadsheet A DSS Tool

• Most widely used tool for decision support
  because it is widely accessible to both
  experienced and inexperienced users
• Features of spreadsheets that support DSS
  construction
• Preprogrammed functions enable the use of many
  mathematical models
• Built in IF-THEN function
• Ability to use Visual Basic for Applications
• Charts, pivot tables, what-if tables, scenario
  management

24
Ethical and Societal IssuesDecisions by Machines

• To what extent should businesses use
  computer-based decisions that can affect
  individuals.
• Mortgage companies, credit card companies,
  employers, banks, etc.
• Distinctions between US and European approaches
• Pros for automated decisions efficiency (cost,
  time, speed)
• Cons injustices due to errors, awareness, and
  finality of decision

25
Model Types

• A model is an abstraction of an object or
  activity (entity)
• Model types
• Physical models like a scale model of a building
• Narrative models provide spoken or written
  descriptions
• Graphic models that use lines, symbols and shapes
  (ER diagram)
• Mathematical models employ equations and(or)
  inequalities

26
Mathematical Models

• A system of mathematical expressions usually
  equations that describe relationships between
  variables in order to better understand or
  describe a real world situation or process.
• Mathematical models usually contain input and
  output variables. Input variables are variables
  that are systematically varied (plugged into the
  equations). The results from the equations are
  the output variables.
• Types of models
• Dynamic versus static
• Stochastic (probabilistic) versus deterministic
• Linear versus nonlinear

27
Sample Models by Functional Area

• Accounting
• breakeven analysis
• tax computations
• depreciation methods
• Finance
• cash flow
• return on investment
• after-tax yield

• Marketing
• product demand forecast
• pricing strategies
• Sales performance
• Production
• product-mix inventory level (EOQ)
• quality control
• production schedule

28
Uses of Models

• Description
• Understanding
• Prediction
• Simulation
• Optimization
• Models typically do more than one of the above