DSS Software Tools

1
DSS Software Tools

• Adopted by
• Dr. Adel S. Aldosary
• for CRP 507, April 27,29, 2003

2
Agenda

• Review Last Lecture
• User-Interface Styles
• DSS Software Categories
• Three Basic Database Structure
• Specialized Modeling Languages
• Video

3
DSS Software Categories

• purchase a turnkey package
• customize a package
• use specialized tools or generators
• custom software
• program from scratch in a suitable language

4
Standard Packages

• help make specific and common decisions
• potential market is large
• decision of financial importance to people
• underlying factors the same for everyone
• e.g., price-earning ratio/chart of companies
• support infrastructure already exists
• e.g., public databases that provide necessary
  data to anyone for a small fee

5
Specialized Tools and Generators

• Standardized building blocks for DSS dev.
• Major categories
• database management packages
• information retrieval (query reporting)
  packages
• specialized modeling packages (including
  spreadsheets) and languages
• statistical data analysis packages
• forecasting packages
• graphics packages

6
Database Management Systems (DBMS)

• a database collection of files
• allow users to store data in an organized form
  and retrieve it on the basis of specified
  selection criteria
• ability to integrate data from several files

7
Database (DB)

• Reflects a conceptual data model
• e.g., Entity-relationship diagram (ERD)
• - specifies the entities about which the database
  contains data and the way in which these entities
  are related

8
Three Basic Database Structures

• Hierarchical DB model
• link records of different types in a strict
  hierarchy from top to bottom
• Network DB model
• provide more flexibility than hierarchical DB
  structure in the way different files are linked
• Relational DBMS
• stores data in the form of 2-dimensional tables
• e.g., Microsoft Access DB4

9
Relational Database Management

• Link records from several tables by matching
  corresponding fields
• Return the set of all records satisfying a given
  set of conditions specified in the retrieval
  operations
• Standardization of interface
• SQL (Structured Query Language)

10
Information Retrieval Packages

• Included as part of a DBMS-based system or sold
  separately
• Overcome difficulties inherent in end-user
  attempts to access a database via SQL
• Strength of a good information retrieval package
• power ability to specify complex queries
• ease of use master the package in a short time
• e.g., graphically oriented info. retrieval
  packages

11
Specialized Modeling Languages

• Packages do not incorporate models themselves,
  but simply make it easier for users to define the
  characteristics of models
• Accounting models a static model with no
  uncertainty
• spreadsheet 2 dimensional grid of cells
• IFPS (Interactive Financial Planning System)
• model exists as a separate, easily visible, and
  easily auditable entity from data

12
Spreadsheets

• Numbers, text, or formulas in cells
• Examples
• built-in functions for mathematical and business
  computations NPV calculations, statistics, etc.
• conditional computation capabilities
• graphs
• extend 2-dimensional grid to 3 or more dimensions
• treat contents of each row as a simple database
• formatting capability

13
Drawbacks of Spreadsheets

• Potential for errors
• Formulas normally hidden from user and scattered
  all over the spreadsheet
• Susceptible to malicious change by users

14
IFPS

• Software developed specifically for financial
  modeling
• Data and models are separate entities
• easier to identify errors
• easier to audit

15
Features of IFPS

• Output cells referred by row and column names
  instead of letter-and-number references
• Formulas apply to entire row (unless otherwise
  stated)
• Backward reference uses the word previous
  instead of obscure cell reference
• Include goal seeking, what-if testing of
  alternatives, and a full set of financial and
  mathematical functions

16
Statistical Data Analysis Packages

• Premise future is in some way tied to factors we
  can change
• Simplest type of prediction
• regression calculations
• develop formula for prediction
• determine correlation (between estimating formula
  and dependent variable) to what degree is the
  dependent variable accounted for by the variables
  considered

17
Common Features of Statistical Packages

• Forecasting methods
• Time-series analysis
• Regression and variance analysis
• Multivariate analysis
• Spectral analysis
• Tools for econometric modeling
• Optimization methods for OM
• Variety of graphical output method
• Extension data access capability for DBMS

18
Forecasting Packages

• Forecasting
• predicting a phenomenon that will take place in
  the future
• e.g., forecast customer order weather forecast
• model how known factors influence other factors
• e.g., economics models
• example of a forecasting problem
• time series forecast

19
Graphing or Charting Packages

• People assimilate data most readily in the form
  of a picture
• Graphs cannot convey differences among numbers
  smaller than 1 percent of their full scale such
  precision usually not required in decision
  support applications
• Many problems/decisions are based on
  trends/differences that are clearly apparent on a
  graph

20
Programming Languages for DSS

• Writing a DSS from scratch
• Third Generation Programming Languages (3GL)
• procedural efficient in run-time
• take a longer time to develop than 4GL
• labor intensive, time consuming, error prone
• Fourth Generation Programming Languages (4GL)

21
Fourth Generation Languages (4GL)

• Specify what the computer is to do, not how the
  computer is to do it
• Trading off flexibility and run-time efficiency
  for speed of development
• Shorter application development by an order of
  magnitude
• Quicker to modify DSS to react to changing
  conditions

22
Why is 4GL not widely used?

• Most existing systems are written in COBOL (3GL)
• More trained and experienced 3GL programmers
• 4GLs are not standardized, unlike 3GLs
• differences among 4GLs
• similarities among 3GLs
• concern with run-time efficiency
• cost of 4GL and its supporting software