DSS DESIGN I

1
DSS DESIGN I

• IS3301 Session 8
• Prof. Mark Nissen

2
Agenda

• Hospital Health Care Services
• DSS Elements Levels
• DSS Design Goals Difficulties
• DSS Design Strategies
• ROMC Example
• Summary

3
Hospital Health Care Services

• JHHS regional health care provider
• 7 facilities, 1000 beds, 3500 employees
• IS 3 of operating budget
• DSS based on SAS
• What capabilities?
• Who uses? What for? What kind of data?
• MAPS?
• DSS developmental approach?

4
DSS Elements
Decision Maker
DGMS
DBMS MBMS
5
DSS Levels Users

• Three levels
• Specific DSS - knowledge worker
• Single application (restaurant advisor)
• DSS generator - modeler/builder
• Family of apps (heuristic source selection)
• RAD tools (LDW, Analytica, VPExpert)
• DSS tools - IT specialist
• Basic tools (VBasic, C, Java, Oracle)
• Few people excel at all three
• DMer often not IT specialist

6
DSS Design Goals

• Support human cognitive tasks involving
  decisions, judgment choice
• Differ from other IS?
• Search engine? Enrollment processing?
• Improve DM efficacy (efficiency?)
• Help structure decision making
• Easy to use, maintain conform to org processes
  DM styles?
• Accurate, reliable trustworthy system?
• Bridge gap between DMer DSS builder?

7
DSS Design Difficulties

• Semi(un)structured problem solving
• Requirements hard to specify ex ante
• DM integral part of DSS
• Wide variation idiosyncrasies
• Must design around DMer
• Human judgment often hard to explain
• Less mature tools methods
• Intelligence hard to capture emulate

8
Informational Characteristics

• Task influences info rqmts (context)
• Inherent accuracy
• Required level of detail
• Time horizon
• Frequency of use
• Internal vs. external source
• Strategic vs. op scope
• Quantifiability
• Currency
• Rqmts also vary with decision maker?

9
Behavioral Characteristics

• Need to understand DMer behavior
• Some behavioral characteristics
• Holistic human DM vs. rationality
• Many (38), diverse DM strategies
• Cognitive limitations DM biases
• Human DM performance degrades - time pressure,
  stress, parallel rqmts
• Problems with aleatory information
• Contingency task structure (S1 F1.7)
• TADMUS?

10
Contingency Task Structure

• Inputs to DSS design
• DM environment
• Internal context?
• External factors?
• Decision maker?
• Task requirements
• DM/processing steps
• DM stress cost
• DM experiential familiarity (KM)
• Specific to DMer, changes thru time

11
Developmental Strategies
Formal DM Staff
End User Development
Development Development
Rigid rqmts Prototyping
Prototyping IS specialists
DM staff
User/DM IT Control Domain
understanding DM direct IT Quality
DM confidence Documentation
DM control High Expertise
Faster development Spec problem
IT
Control Expensive
IT
Quality Time-consuming
Doc
Strong
Weak
12
DSS Design Approach Decision-Maker Centric

• Analyze existing process (not BPR)
• Formalize decision problem
• Decision steps, variables models
• Preliminary KA/CSF analysis
• DM context behavioral description
• ROMC T method
• Knowledge/data acquisition
• Rapid, iterative prototyping

13
ROMC T Method

• DM requirements analysis
• Need process independent approach
• Support DMer idiosyncrasies, contextual
  variations available technologies
• Integrate with all formal decision phases
• Elements on DSS design palette
• Representations, Operations, Memory Aids,
  Controls (ROMC)
• ROMC is technology-neutral
• Technologies

14
Representations

• How DMer views interprets problem
• Can think of user screen displays
• Examples maps, tables, scatterplots charts,
  reports lists, models, rules, natural language
• Key how does DMer represent problem without DSS?
• Parking example?

15
Operations

• Manipulations to data representations
• Help understand, analyze, interpret, etc
• Can think of user actions/analyses
• Examples
• create, retrieve, update, delete info
• plot, scale, mark, color, compare maps/graphs
• generate, edit, check reports/lists
• build, validate, execute, maintain models
• What does DMer do w/o DSS?
• Parking example?

16
Memory Aids

• Assist short- long-term memory
• Can think of data sources
• Examples scratch space, memory, storage,
  reports, scripts, protocols, notes
• Key what aids does DMer use without DSS?
• Parking example?

17
Controls

• How user interfaces with and uses DSS
• Can think of navigation mechanisms
• Examples input forms, menus, question-answer
  pairs, direct manipulation objects, commands,
  scripts, agents, others
• Key match flow of DMer activity without DSS
• Parking example?

18
Technologies

• Technical artifacts that enable ROMC elements to
  be supported via computer
• Can think of IT tools
• Examples imaging, databases, languages, solvers,
  memory, magnetic storage, printers, networks,
  recorders, ES shells, ANNs, GAs, others
• Key what technologies are available, familiar
  low-risk?
• Parking example?

19
ROMC T Palette - Parking
Rep Op Mem Cont Tech

Intel
map
GIS
zoom-in
menus buttons
GIS/DB
mark spot
record spot
?
search
DBMS
list
note rules
sort
Design
table
query
. . .
20
ROMC T Table
Letters (a-z) denote Rs, Os, Ms, Cs Ts
available on DSS designer palette
21
Restaurant Example

• Student volunteer (ate out recently)
• How selected restaurant?
• Understand current process
• Design for improvement via DSS
• Intelligence - need for decision?
• ROMC T
• Design - generate alternatives?
• ROMC T
• Choice - select alternative?
• ROMC T

22
Summary

• DSS three levels, unique goals
• DSS design can be difficult
• Must consider DMer behavior
• DMer-centric design approach
• ROMC T method
• Follow with UML/use cases