Document 7829911

Download Report

Transcript Document 7829911

MCDS methods in strategic
planning- alternatives for AHP
Annika Kangas & Jyrki Kangas
Saariselkä 19.-21.4.2001
Ecological information
• Ecological / recreational information often
has low quality
– risk of ditch maintenance or clearcutting to
watercourse
– wildlife population viability
Need for methods that deal with low quality
information and uncertainty
Public participation
• Public participation (e.g.in State forests)
involves a large number of participants
• Group decision making involves several
DMs
 high costs and poor availability of information
Need for methods that have low information
requirements and enable cheap preference
elicition
Multicriteria approval
• Based on approval voting
– instead of several voters several criteria
considered
• Information requirements
– criteria ranked according to importance
– acceptability of alternatives with respect to each
criteria, for example
• above average  acceptable
• below average  not acceptable
Usability
• Could be used for public participation
– post or internet inquiries
• Criteria values measured in ratio or interval
scale are downscaled to ordinal scale
 information is lost
Outranking
• Ordinal, interval and ratio scale information
can be used
– information transformed to pseudo-criteria
– uncertainty dealt with pseudo-criteria
thresholds
• Weights of criteria interpreted as votes
• If intensities of preferences are known,
information may be lost
Public participation example
• In State owned forests public participation
obligatory
• Case study
– four participants: FPS, regional group, local group and
public
– four main criteria: FPS’s business revenues, socioeconomic values, recreational values and conservational
values, measured with 17 variables
– six strategies
Decision hierarchy
Observed rankings
Strategy
HIPRE Promethee II ELECTRE III
Business
1
1
6
Basic
2
3
1
Forest recreation
3
2
3
Mixed 2
4
4
4
Mixed 1
5
5
2
Nature conservation 6
6
5
Group decision making example
• Jointly owned forests problem in forest
management
– all owners need to approve management actions
• Case study
– three owners with equal share
– 20 forest plans
– six criteria: net incomes, value of the forest, landscape
beauty, blueberry yield, capercaillie viability and
biodiversity
Observed rankings
Alternative AHP Promethee I Promethee II
S1
S2
S3
S4
S5
S6
S7
S8
S9
S10
S11
S12
S13
S14
S15
S16
S17
S18
S19
S20
5.
4.
11.
1.
8.
3.
19.
7.
13.
2.
14.
6.
9.
18.
12.
10.
15.
20.
16.
17.
1.
1.
1.
1.
1.
1.
1.
1.
2.
1.
7.
18.
14.
4.
8.
19.
5.
20.
10.
17.
3.
12.
6.
13.
16.
9.
15.
11.
Promethee II
MA
with AHP weights
7.
4.
3.
1.
10.
12.
.
5.
19.
20.
9.
18.
14.
2.
8.
16.
6.
1.
11.
20.
13.
15.
Requirements
• Methods that utilise both low and high
quality information
– forest information fairly accurate when
compared to ecological criteria
– all information in use, nothing wasted
• Uncertainty dealt with explicitly
– Distributions of uncertain criterion values and /
or criterion weights
SMAA - a possibility
• Stochastic multicriteria acceptance analysis
– what kind of preferences support any one
alternative
• Weight information can be exact, partial or
nonexistent
• Criterion values
– uncertain cardinal values from distribution
– ordinal values converted to cardinal using
simulation