Upload
dillon
View
45
Download
3
Embed Size (px)
DESCRIPTION
Chapter 10 Multicriteria Decision-Marking Models. 投票. 怎樣做會令教學的效果好一點? 用英文? 說話慢一點? 一般用中文,重點用英文? 中文說一片,重點用英文再說一片?. 2. 投票方法. 多輪舉手投票,每輪每人一票 每輪去掉票數最少的選項,直至剩下一選項為止 若兩個或以上的選項票數最少時,只考慮該等選項,仍以多輪舉手投票,每輪每人一票去掉票數最少的選項. 3. 投票結果. 4. 10.3 AHP (Analytical Hierarchy Process) 層級分析法. 5. - PowerPoint PPT Presentation
Citation preview
1
Chapter 10
Multicriteria Decision-Marking Models
2
投票 怎樣做會令教學的效果好一點?
用英文?
說話慢一點?
一般用中文,重點用英文?
中文說一片,重點用英文再說一片?
3
投票方法
多輪舉手投票,每輪每人一票
每輪去掉票數最少的選項,直至剩下一選項為止
若兩個或以上的選項票數最少時,只考慮該等選項,仍以多
輪舉手投票,每輪每人一票去掉票數最少的選項
4
投票結果選項 第一輪 第二輪 第三輪 第四輪 第五輪
用英文?
說話慢一點?
一般用中文,重點用英文?
先用中文,重點再用英文說一片?
選項 第六輪 第七輪 第八輪 第九輪 第十輪
用英文?
說話慢一點?
一般用中文,重點用英文?
先用中文,重點再用英文說一片?
5
10.3 AHP (Analytical Hierarchy Process)
層級分析法
6
AHP (Analytical Hierarchy Process)層級分析法
keys in the scoring method, identifying weights of factors (i.e., objectives) ratings of each alternative for each factor
weight microwave refers stoves
manuf. cap./cost 4 4 3 8market demand 5 8 4 2profit margin 3 6 9 5(long-term) prof./growth 5 3 6 7
transp. costs 2 9 2 4useful life 1 1 5 6
目標選擇 / 選項
7
AHP (Analytical Hierarchy Process)層級分析法
drawback of the scoring or weighting methods:
subjective
hard to simultaneously compare multiple items
questions: Is there any method
with more analytical basis?
easy to compare, e.g., each comparison is only between two options?
weight microwave refers stoves
manuf. cap./cost 4 4 3 8
market demand 5 8 4 2
profit margin 3 6 9 5
(long-term) prof./growth 5 3 6 7
transp. costs 2 9 2 4
useful life 1 1 5 6
8
AHP (Analytical Hierarchy Process)層級分析法
AHP each comparison between two factors or two altrenatives more analytical approach to get the weights
normalized weights of factors normalized priorities of each alternative for factors
Priorities
Alternative 3Alternative 2Alternative 1weightfactor
…………useful life
…………transp. costs
…………(long-term) prof./growth
…………profit margin
…………market demand
…………manuf. cap./cost
sum to 1
sum to 1
sum to 1
.
.
.
9
AHP (Analytical Hierarchy Process)層級分析法
AHP normalized weights of factors normalized priorities of each alternative for factors
Priorities
Alternative 3Alternative 2Alternative 1weightfactor
…………useful life
…………transp. costs
…………(long-term) prof./growth
…………profit margin
…………market demand
…………manuf. cap./cost
question: how to determine those
weights and priorities?
sum to 1
determining the relative importance (i.e., weights) of the factors
sum to 1
determining the relative
importance (i.e., the priorities) of the factors for the alternative
10
AHP (Analytical Hierarchy Process)層級分析法
ideas of AHP to determine the relative importance simple to compare for two alternatives combining the pairwise comparisons into
overall comparisons for weights of all factors, and for priorities of alternatives in each factor
每次都只比較兩樣東西,或是兩目標,或是兩選項,最終將這些兩兩相比轉化成所有選項的比較。
11
Idea of AHP to Determine the Relative Importance
Table 10-1 Preference Scale for the Pairwise Comparisons
Extremely preferred
Very strongly to extremely preferred
Very strongly preferred
Strongly to very strongly preferred
Strongly preferred
Moderately to strongly preferred
Moderately preferred
Equally to moderately preferred
Equally preferred
Numerical ValueVerbal Statement of the preference
9
8
7
6
5
4
3
2
1
12
Idea of AHP to Determine the Relative Importance
simple to compare for two items for relative importance of factors
1/3
1useful life
1Transp. costs
1(long-term) prof./growth
1profit margin
1market demand
1manuf. cap./cost
weightuseful lifetransp. costs(long-term) prof./growth
profit margin
market demand
manuf. cap./cost
3
market demandtransp. costs
row sum reflects the importance of a factor
see their relationship?
…………Options
…………Sound
…………Price
ClarityLuciditySharpweightsFactors
Alternatives
13
Example 10-4
decision: stereo system to purchase brands (i.e., alternatives): Sharp, Lucidity, Clarity criteria (i.e., factors, objectives): sound, price, options
to find the relative importance of criteria to find the relative importance of brands in each criterion
14
Example 10-4: To Find the Relative Importance of Criteria
TotalsOptionsSoundPrice
OptionsSoundPriceCriterion
Table 10-2: Pairwise Comparison Table for the Stereo System Selection Problem
84.3331.583311/31/4311/3431
TotalsOptionsSoundPrice
Average %OptionsSoundPriceCriterion
Table 10-3: Normalized Pairwise Comparison Table for the Stereo System Selection Problem
1.00.15790.2105
1.00.11990.27210.6079
1.00.1250.375
0.5
1.00.07690.23080.69230.6316
…………Options
…………Sound
…………Price
ClarityLuciditySharpweightsFactors
Alternatives
15
Example 10-4: To Find the Relative Importance of Brands in Price
3.33381.75Totals131/2Clarity
1/311/4Lucidity241Sharp
ClarityLuciditySharpCriterion
Table 10-6: Pairwise Comparison Matrix Price
1.01.01.01.0Totals0.32920.300.3750.2857Options0.12260.100.1250.1429Sound0.55710.60.50.5714Price
Average %OptionsSoundPriceCriterion
Table 10-7: Proportion Percentage Matrix for Price
…………Options
…………Sound
…………Price
ClarityLuciditySharpweightsFactors
Alternatives 這步驟是說,以 Price 作
目標,你較喜歡那個選項。留意,喜歡與否,只與喜好有關,沒特別說是較便宜還
是較貴。
16
Example 10-4: To Find the Relative Importance of Brands in Sound
1.58334.57Totals
134Clarity
1/312Lucidity
1/41/21Sharp
ClarityLuciditySharpCriterion
Pairwise Comparison Matrix Sound
1.01.01.01.0Totals0.62320.63160.66670.5714Options0.23950.21050.22220.2857Sound0.13730.15790.11110.1429Price
Average %OptionsSoundPriceCriterion
Proportion Percentage Matrix for Sound
…………Options
…………Sound
…………Price
ClarityLuciditySharpweightsFactors
Alternatives
17
Example 10-4: To Find the Relative Importance of Brands in Options
461.75Totals111/2Clarity111/4Lucidity241Sharp
ClarityLuciditySharpCriterion
Pairwise Comparison Matrix Options
1.01.01.01.0Totals0.23410.250.16670.2857Options0.18650.250.16670.1429Sound0.57940.50.66670.5714Price
Average %OptionsSoundPriceCriterion
Proportion Percentage Matrix for Options
…………Options
…………Sound
…………Price
ClarityLuciditySharpweightsFactors
Alternatives
18
Example 10-4: To Find the Overall Importance of the Brands
overall importance: by weighted score of brands
0.39840.16210.4456Weighted score
0.23410.18650.57940.1199Options
0.62320.23950.13730.2721Sound
0.32020.12260.55710.6079Price
ClarityLuciditySharpweight (%)Criterion
Weights of Factors, Priorities of Brands in Factors, and Weighted Score of Brands
0.1621 = (0.6079)(0.1226)+(0.2721)(0.2395)+(0.1199)(0.1865)
19
Inconsistency in Pairwise Comparison
20
Possibility of Inconsistency in a Pairwise Comparison Matrix
the pairwise comparisons may not be consistent any method to check whether a pairwise comparison
matrix is consistent or not?
A B C
A 1 2 1
B 1/2 1 3
C 1 1/3 1
these pairwise comparisons are not consistent
21
Random Index to Check the Consistency of an AHP
general idea (with detail given later)
consistency index, CI: an index calculated from a pairwise comparison matrix
random index, RI: an index calculated from randomly generated pairwise comparison matrices
The pairwise comparison matrix is inconsistent if CI/RI > some critical value
22
Consistency Index for a Pairwise Comparison Matrix
procedure (for an n-dimensional comparison matrix)
1 Construct a pairwise comparison matrix P
2 Find the normalized weights or priorities for P
3 Calculate = P
4 Calculate ratios for each element of and of , i.e., calculate i/i for i = 1, …, n
5 Calculate average ratio, A = (1/1 + … + n/n)/n
6 Calculate CI = (An)/(n1)
23
Example on Consistency Index
84.3331.5833Totals11/31/4c311/3b431acbaCriterion
1.01.01.01.0Totals0.11990.1250.07690.1579c0.27210.3750.23080.2105b0.60790.50.69230.6316a
Average %cbaCriterion
1 3 4
1/ 3 1 3
1/ 4 1/ 3 1
P
1
0.6079
0.2721
0.1199
2
1 3 4 0.6079 1.904
1/ 3 1 3 0.2721 0.8345
1/ 4 1/ 3 1 0.1199 0.3626
= P 3
1
1
1.9043.13136
0.608=
2
2
0.83453.0669
0.2721=
3
3
0.36263.0242
0.1199=
4
53.13136 3.0669 3.0242
3
average ratio,
3.0742
A
=
6 1
3.0742 33 1
CI=
0.0371
A nn
24
Random Index and the Criterion of Consistency
RI, the consistency index for a pairwise matrix where each pairwise comparison is randomly generated
RI as a function of n in Table 10-5Table 10-5 Random Index Values for the Comparison of n items
consistent if CI/RI < 0.1 consistent because CI = 0.0371, RI = 0.58 for n = 3
25
Final Remarks
Ours is a simplified version of AHP. For example:
AHP is more for a decision problem with hierarchical decisions;
The theory of AHP is related to the maximum eigenvalue and the corresponding eigenvector of a pairwise comparison matrix, something that we have skipped.
26
Chapter 10: Homework for AHP
Problem 16