Analysis representation - PowerPoint PPT Presentation

1 / 12
About This Presentation
Title:

Analysis representation

Description:

Based on a diagnosis and patient age a suitable surgical procedure will be recommended. ... diagnosis. Patient age. Influence diagram: a simple example ... – PowerPoint PPT presentation

Number of Views:41
Avg rating:3.0/5.0
Slides: 13
Provided by: CHMat
Category:

less

Transcript and Presenter's Notes

Title: Analysis representation


1
Analysis representation
  • To document what we know about the decision
    making process ( a more procedural approach) we
    can take techniques from decision analysis.
  • We could use an influence diagram to graphically
    document those factors which may influence a
    decision.

2
The Influence diagram
  • used in decision analysis
  • a graphical tool
  • shows major influences (or variables) on a
    decision
  • see Course Notes p 182 for a reasonably formal
    description of the technique
  • However we will use a far more informal approach
  • Only 3 symbols (icons) are required plus some
    bullet points

Final decision
influence (or variable)
direction of influence
3
Influence diagram a simple example
  • The problem domain is medical. Based on a
    diagnosis and patient age a suitable surgical
    procedure will be recommended.

4
Influence diagram a simple example
  • Outcome 1
  • Outcome 2
  • Outcome 3
  • Outcome 4

Surgical procedure
5
Influence diagram a simple example
diagnosis
Patient age
  • Outcome 1
  • Outcome 2
  • Outcome 3
  • Outcome 4

Surgical procedure
6
Influence diagram a simple example
Blood pressure
Blood test
.etc
  • Juvenile
  • Young
  • Middle aged
  • Old

diagnosis
Patient age
  • Diagnosis A
  • Diagnosis B
  • Diagnosis C
  • Diagnosis D
  • Diagnosis E
  • Outcome 1
  • Outcome 2
  • Outcome 3
  • Outcome 4

Surgical procedure
7
The rule matrix
  • A series of rule matrices can be developed from
    the influence diagram e.g.

8
The rule matrix
  • In this case there are 20 possibilities
  • This matrix documents the final decision
  • Enumerate all or reduce redundancy in this case
    some redundancy has been removed --- a diagnosis
    of C always leads to an outcome of 1 no matter
    the age.
  • Rule matrices can also be developed that lead to
    subgoals' eg

9
Production rules
  • Now a set of production rules can be derived
    from the rule matrices (after the removal of
    redundant matrix entries)
  • For example
  • IF diagnosis Diagnosis A AND age juvenile
    THEN procedure Outcome 3
  • IF diagnosis Diagnosis A AND age young THEN
    procedure Outcome 1
  • IF diagnosis Diagnosis A AND age old THEN
    procedure Outcome 2
  • .
  • .
  • IF diagnosis Diagnosis C THEN procedure
    Outcome 4

10
Production rules
  • Each rule matrix leads to a block (or module) of
    rules
  • Each rule module delivers either a value for the
    final decision or a value for one of the
    influences or variables.
  • So the knowledge base is naturally structured

Rules to give a value for procedure
Rules to give a value for diagnosis
etc
11
Summary
  • identify the decision (goal)
  • identify the direct influences on the
    decision
  • identify subgoals (ie variables or
    influences on the influences above)
  • try and develop several layers of
    subgoals/variables
  • identify possible values for the
    subgoals/variables
  • Work upwards and outwards' until you get to a
    variable which has to asked for.
  • develop a rule matrix from the influence
    diagram
  • remove redundancy
  • develop rules from the matrix

12
Summary
inout
There will eventually be inputs that will lead to
these Influences (subgoals).
inout
subgoal
subgoal
inout
subgoal
subgoal
subgoal
  • Direct influence

Direct influence
Direct influence
Value1 Value 2
Value1 Value 2
Final decision
Value1 Value 2
note all influences, inputs and the decision
node should be accompanied by a list of values
Write a Comment
User Comments (0)
About PowerShow.com