You are on page 1of 12

Method of analysis of hierarchies with the example of choosing

an apartment
CONTENT
1. Method of analysis of hierarchies……………………………3
1.1 The history and essence of it…………………………………..…….3
1.2 The stages of the method…………………………………………….3
2. Example. Choosing an apartment……………….…………..4
1. Method of analysis of hierarchies
1.1 The history and essence of
In 1970, Thomas Saati (USA) developed a method for analyzing hierarchies
(analytical hierarchy process). It belongs to the class of criteria methods. It is
widely used and is still actively used in management practice. Leads the LPR not
to the "right" solution, but to the option that best agrees with its understanding of
the essence of the problem and the requirements for its solution.

1.2 The stages of the method


1. Highlighting the problem. Defining the purpose.
2. Highlighting the main criteria and alternatives.
3. Building a hierarchy: a tree from a goal through criteria to alternatives.
4. Construction of a matrix of pairwise comparisons of criteria by goal and
alternatives by criteria.
5. Application of the method of analysis of the obtained matrices.
6. Determining the weights of alternatives by the hierarchy system.

3
2. Example. Choosing an apartment
1. Purpose: find an apartment for living.
2. Main criterion:
 price;
 size;
 rooms;
 proximity to work;
 home category.
Main alternatives (by ads):
 Apartment 1
 Apartment 2
 Apartment 3

3. Building an alternative tree

An
apartment

Proximity Home
Price Size Rooms
to work category

Apartment 1 Apartment 2 Apartment 3

Scheme 1 - An alternative tree

4. Construction of a matrix of pairwise comparisons


4.1.1 Comparison of criteria.
Initially pairwise comparison on a qualitative scale, with subsequent
conversion to points:
not care, indifferent = 1
slightly better (worse) = 3 (1/3)
4
better (worse) = 5 (1/5)
significantly better (worse) = 7 (1/7)
fundamentally better (worse) = 9 (1/9)
In the intermediate opinion, intermediate points 2, 4, 6, 8 are used.
4.2.1 Drawing up the matrix:
a ij - relation of criterion i to criterion j.

a ij=1 /aij a ii=1

Table 1 - the matrix


price size rooms proximity to work home category
price 1 3 1 1/2 5
size 1/3 1 1/4 1/7 2
rooms 1 4 1 1 6
proximity to work 2 7 1 1 8
home category 1/5 1/2 1/6 1/8 1

4.2 Comparison of alternatives by criteria


Now we make similar matrices of comparison of options (alternatives) for
each criterion.

Table 2 - Comparison of alternatives by price


price Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 4 1/2
Apartment 2 1/4 1 1/5
Apartment 3 2 5 1

Table 3 - Comparison of alternatives by size


size Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1/2 3
Apartment 2 2 1 4
Apartment 3 1/3 1/4 1

Table 4 - Comparison of alternatives by rooms


rooms Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1 2
Apartment 2 1 1 3
Apartment 3 1/2 1/3 1

Table 5 - Comparison of alternatives by proximity to work


proximity to work Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1/3 4
Apartment 2 1 1 5
Apartment 3 1/4 1/5 1

5
Table 6 - Comparison of alternatives by home category
home category Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 2 1/5
Apartment 2 1/2 1 1/6
Apartment 3 5 6 1

5. The method of matrix analysis


5.1.1 Find the sum of the elements of each column.
S j=a1 j+ a2 j +…+ anj

5.1.2 Divide all the elements of the matrix by the sum of the elements of the
corresponding column:
aij
Aij =
Sj

These two actions are called matrix normalization.


Table 7 - the sum of the elements of each column.
a ij price size rooms proximity to work home category
price 1 3 1 1/2 = 0,5 5
size 1/3=0,333 1 1/4=0,25 1/7=0,143 2
rooms 1 4 1 1 6
proximity to work 2 7 1 1 8
home category 1/5=0,2 1/2 = 0,2 1/6=0,167 1/8=0/125 1
SUM 4,533 15,2 3,417 2,768 22

Table 8 - Dividing all the elements of the matrix by the sum of the elements
of the corresponding column:
Aij price size rooms proximity to home category
work
price 1/4,533 3/15,2 1/3,417 0,5/2,768 5/22
size 0,333/4,5333 1/15,2 0,25/3,417 0,143/2,768 2/22
rooms 1/4,5333 4/15,2 1/3,417 1/2,768 6/22
proximity to work 2/4,5333 7/15,2 1/3,417 1/2,768 8/22
home category 0,2/4,5333 0,2/15,2 0,167/3,417 0,125/2,768 1/22

5.1.3 Find the average value for each row:


Aij price size rooms proximity to work home category Averag
price 0,221 0,197 0,293 0,181 0,227 0,224
size 0,073 0,066 0,073 0,052 0,091 0,071
rooms 0,221 0,263 0,293 0,361 0,273 0,282
proximity to work 0,441 0,461 0,293 0,361 0,364 0,384
home category 0,044 0,013 0,049 0,045 0,045 0,039

6
5.1.4 The resulting column sets the"weights" of the criteria in terms of the
goal.
This column is called the goal criteria weight column.
5.1.5 Intermediate output
Weight in fractions Weight as a percentage
price 0,224 22,4%
size 0,071 7,1%
rooms 0,282 28,2%
proximity to work 0,384 38,4%
home category 0,039 3,9%

From the point of view of meeting our goal, the most significant is the
proximity of the apartment to the place of work (38.4%), followed by the number
of rooms (28.2%), then comes the price (22.4%). The size and category of the
apartment have the lowest weight coefficients, totaling only 11%.
Next, 5.2.-5.6. Repeat steps 5.1.1-5.1.5 for the pairwise comparison matrix
by criteria.
5.2. The Criterion Of "Price"
price Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 4 1/2=0,5
Apartment 2 1/4=0,25 1 1/5=0,2
Apartment 3 2 5 1
SUM 3,25 10 1,7

price Apartment 1 Apartment 2 Apartment 3


Apartment 1 1/3,25 4/10 0,5/1,7
Apartment 2 0,25/3,25 1/10 0,2/1,7
Apartment 3 2/3,25 5/10 1/1,7

price Apartment 1 Apartment 2 Apartment 3 Average


Apartment 1 0,308 0,400 0,294 0,334
Apartment 2 0,077 0,100 0,118 0,098
Apartment 3 0,615 0,500 0,588 0,568

price Weight in fractions Weight as a percentage


Apartment 1 0,334 33,4%
Apartment 2 0,098 9,8%
Apartment 3 0,568 56,8%

7
We obtained a vector of weights of objects according to the "Price" criterion.
According to the "price" criterion, the most significant (best, interesting) is
Apartment 3 (56.8%), then Apartment 1 (33.4%), and the least suitable is
Apartment 2 (9.8%). If we chose an apartment only for the price, the choice would
already be obvious.
5.3. The Criterion Of "Size"
size Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1/2=0,5 3
Apartment 2 2 1 4
Apartment 3 1/3=0,33 1/4=0,25 1
SUM 3,33 1,75 8

size Apartment 1 Apartment 2 Apartment 3


Apartment 1 1/3,33 0,5/1,75 3/8
Apartment 2 2/3,33 1/1,75 4/8
Apartment 3 0,33/3,33 0,25/1,75 1/8

size Apartment 1 Apartment 2 Apartment 3 Average


Apartment 1 0,300 0,286 0,375 0,320
Apartment 2 0,600 0,571 0,500 0,557
Apartment 3 0,100 0,143 0,125 0,123

size Weight in fractions Weight as a percentage


Apartment 1 0,320 32%
Apartment 2 0,557 55,7%
Apartment 3 0,123 12.3%

We obtained a vector of weights according to the criterion "size".


According to the "size" criterion, apartment 2 is the most significant
(55.7%), then Apartment 1 (32%), and apartment 3 is the least interesting (12.3%).
5.4. The Criterion Of "Rooms"
rooms Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1 2
Apartment 2 1 1 3
Apartment 3 1/2 1/3 1
SUM 2,5 2,33 6

8
rooms Apartment 1 Apartment 2 Apartment 3
Apartment 1 1/2,5 1/2,33 2/6
Apartment 2 1/2,5 1/2,33 3/6
Apartment 3 0,5/2,5 0,33/2,33 1/6

rooms Apartment 1 Apartment 2 Apartment 3 Average


Apartment 1 0,400 0,429 0,429 0,387
Apartment 2 0,400 0,429 0,429 0,443
Apartment 3 0,200 0,143 0,143 0,170

rooms Weight in fractions Weight as a percentage


Apartment 1 0,387 38,7%
Apartment 2 0,443 44,3%
Apartment 3 0,170 17%
We obtained a vector of weights of objects according to the "Rooms" criterion.
According to the "rooms" criterion, the most significant (best, interesting) is
Apartment 2 (44,3%), then Apartment 1 (38.7%), and the least suitable is
Apartment 3 (17%).
5.5. The Criterion Of "Proximity to work"
proximity to work Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 1/3 4
Apartment 2 3 1 5
Apartment 3 1/4 1/3 1
SUM 4.25 1.53 10

proximity to work Apartment 1 Apartment 2 Apartment 3


Apartment 1 1/4,25 0,33/1,53 4/10
Apartment 2 3/4,25 1/1,53 5/10
Apartment 3 0,25/4,25 0,2/1,53 1/10

proximity to work Apartment 1 Apartment 2 Apartment 3 Average


Apartment 1 0,235 0,217 0,400 0,284
Apartment 2 0,706 0,652 0,500 0,619
Apartment 3 0,059 0,130 0,100 0,096

proximity to work Weight in fractions Weight as a percentage


Apartment 1 0,284 28.4%
Apartment 2 0,619 61,9%
Apartment 3 0,096 9,6%

9
We obtained a vector of weights of objects according to the "P roximity to work"
criterion.
According to the "proximity to work" criterion, the most significant (best,
interesting) is Apartment 2 (61,9%), then Apartment 1 (28,4%), and the least
suitable is Apartment 3 (9,6%).
5.6. The Criterion Of "Home category"
home category Apartment 1 Apartment 2 Apartment 3
Apartment 1 1 2 1/5
Apartment 2 1/2 1 1/6
Apartment 3 5 6 1
SUM 6,5 9 1,37

home category Apartment 1 Apartment 2 Apartment 3


Apartment 1 1/6,5 2/9 0,2/1,37
Apartment 2 0,5/6,5 1/9 0,17/1,37
Apartment 3 5/6,5 6/9 1/1,37

home category Apartment 1 Apartment 2 Apartment 3 Average


Apartment 1 0,154 0,222 0,146 0,174
Apartment 2 0,077 0,111 0,122 0,103
Apartment 3 0,769 0,667 0,732 0,723

home category Weight in fractions Weight as a percentage


Apartment 1 0,174 17,4%
Apartment 2 0,103 10,3%
Apartment 3 0,723 72,3%

We obtained a vector of weights of objects according to the "Home


category" criterion.
According to the "home category" criterion, the most significant (best,
interesting) is Apartment 3 (72,3%), then Apartment 1 (17,4%), and the least
suitable is Apartment 2 (10,3%).
6. Determining the weights of alternatives
As a result of point 5 formed:
the vector of weights of criteria;
matrix of weights of alternatives for each criterion (consisting of the
resulting weight columns).
10
Weight in fractions
Price 0,224
Size 0,071
Rooms 0,282
Proximity to work 0,384
Home category 0,039

Price Size Rooms Proximity to work Home category


Apartment 0,334 0,320 0,387 0,284 0,174
1
Apartment 0,098 0,557 0,443 0,619 0,103
2
Apartment 0,568 0,123 0,170 0,096 0,723
3

Multiplying the resulting matrix by a column by the rule row*column


(matrix), we get the weight of alternatives in terms of achieving the goal:
0,224

(
0,334 0,320 0,387 0,284 0,174 0,071
) ( )(
0,323
0,098 0,557 0,443 0,619 0,103 X 0,282 = 0,428
0,568 0,123 0,170 0,096 0,723 0,384
0,039
0,249 )
As a result, we get the weight of alternatives in terms of achieving the goal:
Weight in fractions Weight as a percentage
Apartment 1 0,323 32,3%
Apartment 2 0,428 42,8%
Apartment 3 0,249 24,9%

Thus, Apartment 2 is the most attractive for our purpose.

11
12

You might also like