Tutorial_4_Changing from AHP to ANP thinking.ppt

8/15/2019 Tutorial_4_Changing from AHP to ANP thinking.ppt

http://slidepdf.com/reader/full/tutorial4changing-from-ahp-to-anp-thinkingppt 1/16

The Analytic Network Process

Decision Making with Dependence and Feedback



Changing from a Hierarchical

Model to a Network Model



Network Models do not have

Goals

A network model has a

criteria cluster and an

alternatives cluster, but

no goal.

Step 1. Compare the

alternatives with respect

to each criterion (like we

do in AHP models).



Establishing Priorities for Criteria

In AHP one answers the question: “Howimportant is Criterion1 compared to

Criterion2 with respect to the goal?”Sometimes too abstract a question to answer.

In ANP criteria are prioritied !" as#ing

how important the" are in the alternati$es

!eing considered% Usually an easier question to

answer when looking at actual alternatives that have the

properties in question.



Cars compared with respect to Prestige

(and similarly for the other criteria)

.707

.070

.!



Feedback: Link Alternatives to Criteria and

compare for Preference

For eample! a pairwise "#estion wo#ld be: $For the Ac#ra %L! which do

yo# like better! its prestige or its price&' $ts prestige or its *P+&' etc,

."##

.0$$

.07#

.!%%



*ake Comparisons

&or each car pairwise compare thecriteria 'or importance to esta!lish the

priorities o' the criteria in that car:

Prestige( Price( )P*( and Com'ort+his results in a $ector o' priorities o'

the criteria 'or each car%



Pairwise Compare Criteria for each Car

,esults 'or Acura +-

,esults 'or +o"ota Camr"

Acura +-

+o"ota Camr"



Pairwise Compare Criteria wrt Cars

,esults 'or Honda Ci$icHonda Ci$ic



+he .nweighted /upermatri0

&or e'am(le) the (riorities in the last col*mn come fromcom(aring the criteria with res(ect to the Honda Civic. That

is) what we val*e most a+o*t the Honda is its (rice.



+he limit supermatri0 is o!tained !" raising the weighted supermatri0 to powers

until it con$erges in this case all columns are the same( though not alwa"s true%

-imit /upermatri0

A-P A.P

Ac#ra %L ,/01 ,2//

%oyota Camry ,312 ,455

.onda Civic ,267 ,/00

+he 'inal answer( the

s"nthesis( !elow is

o!tained !" normaliing

the raw $alues 'or the

alternati$es 'rom the limitsupermatri0:



8hy are the res#lts different&

3n A.P 4 the user( going top5down ma#ing

comparisons( when as#ed without re'erring

to actual alternati$es( o$er5estimated the

importance o' cost%

3n A-P 4 the user learned through

'eed!ac# comparisons that his6her priorit"

'or Cost is not nearl" as high as originall"

thought when as#ed the question a!stractl"(

while Prestige gets more weight%

This is why a husband and wife, setting out to the dealership to buy a “sensible” low

priced car, walked out with something entirely different and much more costly.

When they saw the shiny cars on the showroom floor they reised their priorities.



nner and 9#ter ependence

39#ter dependence 5 the parent node andthe nodes to !e compared are in di''erent

clusters% A directed lin# appears from the

parent node cluster to the other cluster%

3nner dependence 5 the parent node and

the nodes to !e compared are in the same

cluster% +he cluster is lin#ed to itsel' and aloop lin# appears%



7e can add some inner dependence: eg( price

is in'luenced !" prestige and com'ort8 and

prestige is in'luenced !" price and com'ort%



An 90ample o' an Inner ependent

Comparison is shown !elow:

,hich infl*ences Price more) Prestige or Comfort-

,hich infl*ences Prestige more) Price or Comfort-



+he results with the interdependencies in

place are:

A-P A-P A.P

(8ith inner (-o inner

dependencies) dependence)

Ac#ra %L ,/;1 ,/01 ,2//

%oyota Camry ,3;4 ,312 ,455

.onda Civic ,223 ,267 ,/00

Tutorial_4_Changing from AHP to ANP thinking.ppt

Documents