You are on page 1of 14

Assignment of Weights

• Other methods, besides arbitrary, for weight


assignment exist

• There are both direct and indirect weight


elicitation techniques

Source: The Engineering Design of Systems


Models and Methods
By Dennis M. Buede
(Chapter 13)
Alternative Weighting Methods
Direct Weight Elicitation Techniques

– Rank-Order Centroid Technique

1. List Objectives in order from most important


to least important
2. Use one of the following formulas for
assigning weights
Alternative Weighting Methods
K  ri  1
1. Rank Sum: wti  K
• ri is the rank of the ith objective
• K is the total number of objectives
K r
j 1
j 1

z
( K  ri  1)
2. Rank Exponent : wti  K
• ri is the rank of the ith objective
• K is the total number of objectives
 (K  r
j 1
j  1) z

• z is an undefined measure of the dispersion


in the weights
Alternative Weighting Methods
3. Rank Reciprocal: 1 / ri
wti  K
• ri is the rank of the ith objective
• K is the total number of objectives  (1 / r )
j 1
j

4. Rank Order  1  K  1 
Centroid: wt i      

 K  j 1  r j 
 1 1 1 
wt1  1    ... / K
 2 3 K
 1 
wt K   0  0  0  ... / K
 K
Alternative Weighting Methods
Indirect Weight Elicitation Techniques

Trade Offs:
• Objectives ranked in order of their overall swing in value
• Stakeholders asked if overall swing weight of the second
objective is as great as the swing from the lowest to some
intermediate point of the value scale of the first objective.
• The third objective is can now be compared to intermediate points
on either the first or second ranked objective and so on…
• Method works well when the value curves are firmly established
and when the value curves are continuous
Alternative Weighting Methods
Analytical Hierarchy Process:

1. Stakeholders asked to compare objectives two at a time


• Comparisons made using a numerical scale that ranges from 9 times
more valuable to one ninth as valuable (an equivalent verbal scale
can also be used)
• If there are K objectives then K(K-1)/2 comparison questions would
be asked
2. The responses are input into a matrix upon which an eigenvector
calculation is performed
3. The eigenvector is normalized thus returning the determined
weights
Alternative Weighting Methods
Balance Beam Approach (This method was used to determine the weights)
1. Stakeholders establish a rank order of the overall swing weights of the
objectives.
2. A series of questions is posed beginning with “Is the overall swing in the value
of the first objective (a) greater, (b) less than, or (c) equal to the overall swing
in values of the second and third objectives
3. If answer is “less than” then the third objective is dropped and replaced by
the fourth objective. (If “greater than” then the fourth objective is added to the
second and third )
• The goal is to establish a series of equations that define the swing weights
for all of the objectives
4. The least valued objective is given a weight of 1 and then the stake holders
are asked to assign a swing weight to the second least weighted objected
then this information is used to solve the system of equations
5. The results are then normalized into weights with values between 0 and 1
Example: Bicycle Wheel
• The company 322 Bikes decides to design
a new rear wheel due to high failure rates
with the current design.

• Three representatives set out to interview


potential customers and returned with that
following results…
Example: Bicycle Wheel
• Marketing Representative A:
– “The most important feature that customers require is
that the product must sustain wear and tear for a long
period of time.”
• Marketing Representative B:
– “Also, our customers want a strong and affordable
item.”
• Marketing Representative C:
– “They want a strong wheel but they also don’t want it
to weight a lot.”
Example: Bicycle Wheel
• From the customer needs, the design
department of 322 Bikes reduced the
customer needs into three key concepts:
– Long Product Lifetime
– Low Weight
– Low Cost

This example will be carried out through the rest of the semester. It will be used to illustrate the use of
Decision Making under Uncertainty.
Example: Bicycle Wheel
• A good design will be judged on the
performance criteria (Evaluation Measures)
which are:
– Product life time
– Weight
– Cost
• First, we arrange these evaluation measures
in order of importance:
1. Cost: Consumer are initially concerned with price.
2. Weight: More discerning consumer are also
concerned with the weight.
3. Lifetime: Lifetime is important but consumers
assume that the wheel will last a reasonable
amount of time.
Example: Bicycle Wheel
Rank Exponent Rank Order
EM Rank Sum Rank Reciprocal
(dispersion=0.2) Centroid

Cost 0.50 0.37 0.55 0.61

Weight 0.33 0.34 0.27 0.28

Lifetime 0.17 0.29 0.18 0.11

The above are the relative weights calculated using the different direct
weighing techniques.
Note that, in each column, the total weights add up to 1. The relative
importance of each Evaluation Measure is denoted by what
percentage of the whole each EM occupies. The higher the value, the
more important the EM.
Example: Bicycle Wheel
• Balance Beam Approach (Indirect Method)
• System of equation are as follows:
Cost  Weight  Lifetime

Cost  Weight
Weight  Lifetime

• The smallest EM, Lifetime’s weight, is set to 1.
• From the equations, the other weights are chosen as
follows:
– Weight = 1.2
– Cost = 1.5 (this means that Cost is 1.5 times as important as
Lifetime.)
Example: Bicycle Wheel
• The normalized weights are then:
1.5
Cost   0.41
1  1.2  1.5
1.2
Weight   0.32
1  1.2  1.5
1
Lifetime   0.27
1  1.2  1.5

Please note: All of the above methods are implemented in an Excel


macro. If you are unsure of which method to use, it is easy to perform a
few different weight calculations and then choose the results that best
match your preferences.

You might also like