Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Rajnish Kumar
Criteria 1
Alt 1
Criteria 2
Alt 2
Alt 3
Criteria 3
Alt 4
Price in Rs
lakh
Mileage in
km/l of
fuel
Comfort
on a scale
of 10
850000
12
900000
11
1000000
10
It is certain that the three criteria will have different weights or priority
for different people. Now, pair wise comparison has to be made for the
criteria. Pair wise comparison is comparing the criterion to another for it
respective importance vis--vis the other.
This scale may explain it better.
Price
Mileage
Price
Comfort
Mileage
Comfort
Equal
Moderate
Strong
Very Strong
Extreme
So, if you mark 3 towards the price on the pair wise comparison of price
and mileage, it would mean price is preferred moderately over mileage.
Next step is to form a matrix of the pair wise comparisons.
Price
Mileage
Comfort
Price
Mileage
1/3
1/3
Comfort
1/5
Lets see how this matrix is formed. Since we have three attributes
to be compared a 3 by 3 matrix is formed. The diagonal elements of
the matrix are obviously 1. The upper row has to be filled. The rules
are:
4
In this case, the attribute, price is 3 points higher than mileage and
5 points higher for comfort. To fill the lower triangular matrix, we
use the reciprocal values of the comparisons.
Without going into the mathematics of matrix, the AHP process will
be explained to let the reader have a fair idea of this wonderful
technique, which has established itself as the prime decision making
technique in management science.
We sum each column of the matrix to get
Price
Mileage
Comfort
Price
Mileage
1/3
1/3
Comfort
1/5
Sum
23/15
19/3
Now we divide each element of the matrix with the sum of its
column, to get the normalized relative weight. The sum of each
column should be 1.
5
Price
Mileage
Comfort
Price
15/23
3/7
15/19
Mileage
5/23
1/7
1/19
Comfort
3/23
3/7
3/19
Sum
Mileage
Comfort
Average
or
Priority
Vector
Price
0.6522
0.4286
0.7895
0.6234
Mileage
0.2174
0.1429
0.0526
0.1376
Comfort
0.1304
0.4286
0.1579
0.2390
Sum
But an important point here, how consistent the comparison is. For
the consistency check, we have to calculate the Principal Eigen
6
Mileage
Comfort
Priority
Vector
Price
(P)
Mileage
(M)
Comfort
(C)
0.6234
1/3
1/3
0.1376
1/5
0.2390
Sum
23/15
19/3
Calculation
of Principal
Eigen
Vector
23/15 x 0.6234
7 x 0.1376
19/3 x 0.2390
Adding up
3.43
3.43 3
=
1
31
=0.215
10
RI
0.58
0.9
1.12
1.24
1.32
1.41
1.45
1.49
Using RI the Consistency Ratio (CR) is found out, which must be less
than 10% for decision to be termed consistent. CR is a comparison
between Consistency Index and Random Consistency Index, or in
formula
=
8
Mileage
Comfort
Price
Mileage
1/3
Comfort
1/5
1/3
Sum
23/15
13/3
Mileage vs comfort has been revised and mileage has been placed at
higher preference, repeating all steps as done before, the final
matrix and principal eigen vector calculations are,
9
Price
Mileage
Comfort
Priority
Vector
Price
0.65217
0.69231
0.55556
0.6333
Mileage
0.21739
0.23077
0.33333
0.2605
Comfort
0.13043
0.07692
0.11111
0.1062
Sum
1.5333
4.3333
9.0000
max=
3.055
Price in Rs
lakh
Mileage in
km/l of
fuel
Comfort
on a scale
of 10
850000
12
900000
11
1000000
10
10
1. For higher the better, i.e. positive criteria, take the highest
value as denominator and divide all elements of the column
by it.
2. For lower the better, i.e. negative criteria, take the highest
value as numerator and element as denominator for each cell
of the respective column.
Mileage
Comfort
1.18
(1000000/850000)
1.00
(12/12)
0.60
(6/10)
1.11
(1000000/900000)
0.92
(11/12)
0.80
(8/10)
1.00
(1000000/1000000)
0.75
(9/12)
1.00
(10/10)
Using the weights computed from AHP, the score for each car will is
calculated and cars ranked in that order.
The weight of each criterion is multiplied with the cell value of row
for each alternative and all these values are added up to get the final
score. The alternative with the highest score is the choice.
11
Total
Score
Price
Mileage
Comfort
0.6333
Weights
0.2605
0.1062
0.7451
0.2605
0.0637
1.0693
0.7037
0.2388
0.0850
1.0274
0.6333
0.1954
0.1062
0.9349
BEST
Saaty, T.L., 2001. Decision Making for Leaders: The Analytic Hierarchy Process
for Decisions in a Complex World, (Analytic Hierarchy Process Series, Vol. 2),
RWS Publication, Pittsburgh.
Bhushan, N., Rai, K., 2004. Strategic Decision Making- Applying the Analytic
Hierarchy Process, Series: Decision Engineering, Springer India.
http://mi.boku.ac.at/ahp/ahptutorial.pdf
http://www.business.pitt.edu/faculty/papers/saaty-into-to-ahpmathematics.pdf
http://www.dii.unisi.it/~mocenni/Note_AHP.pdf
http://analytics.ncsu.edu/sesug/2012/SD-04.pdf
http://giswin.geo.tsukuba.ac.jp/sis/gis_seminar/How%20to%20do%20AHP%20
analysis%20in%20Excel.pdf
Contact:
rajnishkumar@nair.railnet.gov.in
rajnishkumar1971@gmail.com
13