PROBLEMA 1.

Decisions under a risk environment

A company dedicated to manufacturing different turned parts must decide whether
buys it from an outside supplier. The profits depend on the demand of the product.

Table 1. Decision alternatives according to the states of nature

States of nature

Decision alternative Demand low average Demand High Medium

Demand low-utility
- utility - utility

Manufacture 35 37 38
Subcontract 33 35 36
Buy 38 40 41
Lease 40 42 43
Probabilities Ʃ = 1 0.23 0.21 0.25

PART 2. DECISION TREES, EVPI and EVMI

Preguntas

a.   Use EVPI to determine if the company should try to get a better estimate of the

b. A test market study of potential product demand is expected to report a favorab

favorable unfavorable
P(F/low) = 0,3 P(D/low) = 0,7
P(F/low average) = 0,36 P(D/ low average) = 0,64
P(F/high medium) = 0,28 P(D/ high medium) = 0,72
P(F/high) = 0,5 P(D/high) = 0,5

c.   What is the expected value of market research information?

d.   What is the efficiency of the information?

SOLUTION

Paso 1. Decision trees

DECISION PROBABILITY

Low (0.23)
 Low average (0.21)
Manufacture
Node 2
High medium(0.25)

High (0.31)

Low (0.23)

Low average (0.21)

Subcontract
Node 3
High medium(0.25)

High (0.31)
Node 1

Low (0.23)

Low average (0.21)

Buy
Node 4
High medium(0.25)

High (0.31)

Low (0.23)

Low average (0.21)

Lease
Node 5
High medium(0.25)

High (0.31)

Paso 2. Expected value without perfect information - EVwoPI

Node 2 37.72
Node 3 35.72
Node 4 40.72
 Node 5 42.41

Node 1 = Max (Node 2, Node 3, Node 4, Node 5)

Node 1= 42.41

The decision recommended by the method of the expected value is subcontract with a payment of 42,41 million

If we remember,

EVPI = |EVwPI - EVwoPI|

EVwPI = (0,31*44)+(0,25*43)+(0,21*42)+(0,23*40) = 42,41 42.41

EVPI = 42.41

EVPI = 0

The expected value wit perfect information is the 0 million

Paso 3. we apply the Bayers theorem to find the posterior porbabilities

favorable
P(F/low) = 0,3
P(F/low average) = 0,36
P(F/high medium) = 0,28
P(F/high) = 0,5

𝑃(𝑠_1 )= 0,23 𝑃(𝑠_2 )= 0,21 𝑃(𝑠_3 )= 0,25 𝑃(𝑠_4 )= 0,31

favorable
P(F/low) = 0,3
P(F/low average) = 0,36
P(F/high medium) = 0,28
P(F/high) = 0,5

Favorable

Previous probabilities
State of nature Conditional
P(sj)
probabilities P(D/sj)
Low 0.23 0.3
Low average 0.21 0.36
 High medium 0.25 0.28
High 0.31 0.5
𝑃(𝑠_1 )= 0,23 𝑃(𝑠_2 )= 0,21 𝑃(𝑠_3 )= 0,25 𝑃(𝑠_4 )= 0,31 P(F)

unfavorable
P(D/low) = 0,7
P(D/ low average) = 0,64
P(D/ high medium) = 0,72
P(D/high) = 0,5

Unfavorable

Previous probabilities
State of nature P(sj) Conditional
probabilities P(D/sj)

Low 0.23 0.7

Low average 0.21 0.64
High medium 0.25 0.72
High 0.31 0.5
P(F)

Paso 4. Decision trees

PROBABILITIES DECISION

Manufacture

FAVORABLE 0.37
Node 2

Subcontract
 Buy

Lease
=(42,65∗0

Node 1

Manufacture

Subcontract
 UNFAVORABLE 0.63 Buy
Node 3

Lease

Paso 5. Expected value sample information.- EVMI

Node value 1 = 42.68

to calculate the Expected value sample information we have to

EVMI = |EVwMI - EVwoMI|

EVwMI = 42.68

EVwoMI = 42.41

EVMI =
𝐸=𝐸𝑉𝑀𝐼/𝐸𝑉𝑃𝐼 𝑋100" " 0.27
EVPI = 0

Eficiencia 𝑋100" "

𝐸=0,27/0

R//: Se presenta una indeterminacion (error), no se puede brindar solución.

According to the corresponding information in Table 1 and the Predicted Value of Perfect Information (EV
Value of Sample Information (EVMI) and Decision Trees, respond: 
a. Use EVPI to determine if the company should try to get a better estimate of the demand.

 R//: El valor esperado de la información perfecta (EVPI) es de 0 millones donde, el valor esperado sin info
igual al valor esperado con información perfecta por lo cual la estimación de la demanda en nula (0) 

b. A test market study of potential product demand is expected to report a favorable (F) or unfavorable (U
relevant conditional probabilities are: 
According to the corresponding information in Table 1 and the Predicted Value of Perfect Information (EV
Value of Sample Information (EVMI) and Decision Trees, respond: 
a. Use EVPI to determine if the company should try to get a better estimate of the demand.
 R//: El valor esperado de la información perfecta (EVPI) es de 0 millones donde, el valor esperado sin info
igual al valor esperado con información perfecta por lo cual la estimación de la demanda en nula (0) 
b. A test market study of potential product demand is expected to report a favorable (F) or unfavorable (U
relevant conditional probabilities are: 

R//: El estudio de Mercado de prueba de la demanda potencial de producto reporta una des favorabilida
posterior de 0,63

c. What is the expected value of market research information?

 R//: El valor esperado de la de la información de investigación de mercado es fabricación con un pago de
d.  What is the efficiency of the information?
R//: De acuerdo con la información asociada al estudio de mercado no se puede obtener una eficiencia p
presenta una indeterminación y no se puede calcular.
 sk environment:
st decide whether to manufacture a new product at its main plant, or if it
nd of the product. Table 10 shows projected profits, in millions of pesos.

nature

Demand High - utility

40
38
43
44
0.31

r estimate of the demand.

o report a favorable (F) or unfavorable (U) condition. The relevant conditional probabilities are:

unfavorable
= 0,7
verage) = 0,64
medium) = 0,72
= 0,5

DEMAND

35 8.05
37 7.77
37.72
38 9.5

40 12.4

33 7.59

35 7.35
35.72
36 9

38 11.78

38 8.74

40 8.4
40.72
41 10.25

43 13.33

40 9.2

42 8.82
42.41
43 10.75

44 13.64
ayment of 42,41 million

million

unfavorable
P(D/low) = 0,7
P(D/ low average) = 0,64
P(D/ high medium) = 0,72
P(D/high) = 0,5

able

Joint probabilities. P(Dnsj) Later probabilities.

P(sj/D)
0.069 0.19
0.0756 0.20
 0.07 0.19
0.155 0.42
0.3696

rable

Joint probabilities. P(Dnsj) Later probabilities.

P(sj/D)

0.161 0.26
0.1344 0.21
0.18 0.29
0.155 0.25
0.6304

PROBABILITY DEMAND

Low (0.19) 35

Low average (0.20) 37

38.07
Node 4
High medium(0.19) 38

High (0.42) 40

Low (0.19) 33

Low average (0.20) 35

36.07
Node 5
High medium(0.19) 36
 High (0.42) 38

Low (0.19) 38

Low average (0.20) 40

41.07
Node 6
High medium(0.19) 41

High (0.42) 43

Low (0.19) 40

Low average (0.20) 42

42.65
Node 7
=(42,65∗0,37)+(42,69∗0,63)
High medium(0.19) 43

High (0.42) 44

Low (0.26) 35

Low average (0.21) 37

37.89
Node 8
High medium(0.29) 38

High (0.25) 40

Low (0.23) 33

Low average (0.25) 35

35.87
Node 9
High medium(0.27) 36

High (0.24) 38

Low (0.23) 38

Low average (0.25) 40

40.92
 40.92
Node 10
High medium(0.27) 41

High (0.24) 43

Low (0.23) 40

Low average (0.25) 42

42.69
Node 11
High medium(0.27) 43

High (0.24) 44

erfect Information (EVPI) theory, the Expected

demand.

valor esperado sin información perfecta es

manda en nula (0) 

e (F) or unfavorable (U) condition. The

 erfect Information (EVPI) theory, the Expected

demand.
valor esperado sin información perfecta es
manda en nula (0) 
e (F) or unfavorable (U) condition. The

a una des favorabilidad con una probabilidad

cación con un pago de $ 42,41 millones.

tener una eficiencia porque el resultado

l probabilities are:
42.65

42.675216

42.69
 SITUACION DE INCERTIDUMBRE
PROBLEMA 2
The company is thinking of acquiring machinery with new technology to carry out its workshop w
The purchase will be decided according to several alternatives presented by the seller (adaptabil
to facilitate the implementation in the workshop. The decision variables presented below represe
cost of adaptation that will arise after acquiring the machinery and training the workers in their
Table 11 shows the costs in millions of currency units per technology.

Table 2. Uncertainty adaptation new technology

Event

Does not Fits Fits

Alternative Fits well
fit acceptably successfully

Technology 1 780 810 818 860

Technology 2 880 820 855 820
Technology 3 830 875 878 900
Technology 4 630 872 812 910

Criteria
LAPLACE

1/4 1/4 1/4 1/4

Alternative Does not Fits Fits VM
Fits well
fit acceptably successfully
Technology 1 780 810 818 860 817
Technology 2 880 820 855 820 843.75
Technology 3 830 875 878 900 870.75
Technology 4 630 872 812 910 806

Therefore, the lowest level of supply is associated with the lowest expected cost (806) specified technology 4.

Wald or pessimistic

Does not Fits Fits

Alternative Fits well MAX{Xy}
fit acceptably successfully

Technology 1 780 810 818 860 860

Technology 2 880 820 855 820 880
Technology 3 830 875 878 900 900
Technology 4 630 872 812 910 910

Hurwicz
coefficient optimist a 0.7
coefficient pesSimistic (1 - a): 0.3

A1= 860*0,7 + 780*0,3 = 836

A2= 880*0,7 + 820*0,3 = 862
A3= 900*0,7 + 830*0,3 = 879
A4= 910*0,7 + 630*0,3 = 826

Valor esperado (A)= (max * a + min * a)

Nivel de
Does not Fits Fits
Alternative Fits well seguridad
fit acceptably successfully Max {Xy}
Technology 1 780 810 818 860 860
Technology 2 880 820 855 820 880
Technology 3 830 875 878 900 900
Technology 4 630 872 812 910 910

Nivel de
Does not Fits Fits
Alternative Fits well seguridad
fit acceptably successfully Max {Xy}
Technology 1 780 810 818 860 860
Technology 2 880 820 855 820 880
Technology 3 830 875 878 900 900
Technology 4 630 872 812 910 910

When dealing with costs, the alternative associated with the lowest weighted average is selected. The opti

Savage

Does not Fits Fits

Alternative Fits well
fit acceptably successfully

Technology 1 780 810 818 860

Technology 2 880 820 855 820
Technology 3 830 875 878 900
Technology 4 630 872 812 910

Does not Fits Fits

Alternative Fits well
fit acceptably successfully
Max {Py}
(780 - ( 818 - ( 860 -
Technology 1 (810 - 810)=0
630)=150 812)=6 820)=40 150
(880 - (820 - ( 855 - (820 -
Technology 2
630))=250 810)=10 812)=43 820)=0 250
 (830 - ( 875 - ( 878 - ( 900 -
Technology 3
630))=200 810)=65 812)=66 820)=80 200
(630 - ( 872 - (812 - (910 -
Technology 4
630))=0 810)=62 812)=0 820)=90 90

Matrix of benefits

Determine the optimal size of the premises to be purchased, using the methods of
LAPLACE, WALD, HURWICZ AND SAVAGE. Hurwicz Alpha 0,70.

R//: According to the criteria of costs of the place the optimal decision level is
technology 4, with $ 806 dollars. with criterion the Wald or pessimistic, the optimal
decision level is technology 1,. with $ 860,00 dollars. with criteria the Hurwicz, the
optimal decision level is technology 4, with $ 826,00 dollars and with criteria the
Savage (Cost matrix), the optimal decision level is technology 4, with $ 90 dollars.
arry out its workshop work.
by the seller (adaptability), this
resented below represent the
g the workers in their use.

lower cost

fied technology 4.

MINI MAX
 Nivel
optimista Media
Min {Xy} ponderada
780 836.00
820 862.00
830 879.00
630 826.00

Nivel
optimista Media
Min {Xy} ponderada
780 836.00
820 862.00
830 879.00
630 826.00

average is selected. The optimal solution is given by technology 4.

 Problem 3. Decision in uncertainty:

PLAYER B MAXIMIN
65 82 72 65 68 65
PLAYER A 78 89 56 89 81 56
92 86 83 64 72 64
89 88 76 67 75 67
67 59 89 65 79 59
MINIMAX 92 89 89 89 81 There is no established cha

STEPS SOLVER
Strategy maximin = minimax

Q1 Q2 Q3 Q4 Q5
0.114415 0 0.39495798 0.49062702 0

PLAYER B
P1 0 65 82 72 65 68
PLAYER A

P2 0.409825 78 89 56 89 81
P3 0.12799 92 86 83 64 72
P4 0 89 88 76 67 75
0.462185 67 59 89 65 79
Suma 1
Maximun 92 89 89 95 81

Vexpected = 74.707822 74.7504848 74.7078216 74.7078216 78.9237233

MaxZ = V 74.707822

The maximin and the minimax would already be the same and this would be the saddle point = 74,707822

R//: The maximin and the minimax would already be the same and this would be the saddle point = 74,70782
here is no established chair point

Suma
1

Minimax Vexpected = MinZ = V

65 67.7647059 74.7078216
56 74.7078216
64 74.7078216
67 73.0717518
59 74.7078216

ddle point = 74,707822

e the saddle point = 74,707822

 Problem 4. Decision in uncertainty:

In order to determine the decision conditions in the market, the Game Theory will be
used, using the graphical solution of the type (2 x N) and (2 x M) to estimate the
strategy and value of the game for the following data:

Table 4. Data for matrix strategy mxn

Player 2
Strategy
A B
I 5 6
Player 1
II 4 3

For column the player

Strategy A Strategy B

P1+P2=1 P1+P2=1

Vexpected = 5P1 + 4P2 Como P1 + P2 =1 P2= 1 - P1 Vexpected = 6P1 + 3P2

Vexpected = 5P1 + 4(1-P1) Replace P2 Vexpected = 6P1 + 3(1-P1)
Vexpected = 5P1 + 4 - 4P1 Vexpected = 6P1 + 3 -3P1
Vexpected = 1P1 + 4 Vexpected = 3P1 + 3

Strategy A Strategy B
Si P1 = 1 Ve = 5 Si P1 = 1 Ve = 6
Si P1 = 0 Ve = 4 Si P1= 0 Ve = 3

12_ Then A and B

Vexpected = 3P1 +3
9_ Vexpected = 1P1 + 4

3P1 + 3 = 1P1 + 4
6_ A 3P1 - 1P1 = 4-3
2P1 = 1
P1 = 1/2
3_ B P1 = 0.5
 P2 = 1 - P1 = 1 - 1/2
0_ P2 = 0.5
1
VALUE OF THE GAME

Vexpected = + 3 =
Vexpected = 4.5

This means that if both players determine the probabilities of the actions
of each of their strategies, both will have the possibility of 2,5. This
represents a balanced game.

For the player row

Player 2
Strategy
A B
I 5 6
Player 1
II 4 3

Strategy 1 Strategy 2

Q1+Q2=1 Q1+Q2=1

Vexpected = 5Q1 + 6Q2 Como Q1 + Q2 =1 Q2= 1 - Q1 Vexpected = 4Q1 + 3Q2

Vexpected = 5Q1 + 6(1-Q1) Vexpected = 4Q1 + 3(1-Q1)
Vexpected = 5Q1 +6 - 6Q1 Vexpected = 4Q1 + 3 - 3Q1
Vexpected = -1Q1 + 6 Vexpected = 1Q1 + 3

Strategy 1 Strategy 2
Si Q1 = 1 Ve = 5 Si Q1 = 1 Ve = 4
Si Q1 = 0 Ve = 6 Si Q1= 0 Ve = 3

12_

Then A and B
9_
Vexpected = -1Q1 + 6
Vexpected = 1Q1 + 3
6_
 1 -1Q1 + 6 = 1Q1 + 3
-1Q1+1Q1 3 - 6
3_ 2 0Q1 -3
Q1 = 0/-3
Q1 = 0
0_
1 Q2 = 1 - Q1 = 1 - 0/3
Q2 = 1

VALUE OF THE GAME

Vexpected =Q1 + 6
Vexpected = 6

This means that if both players determine the probabilities of the actions
of each of their strategies, both will have the possibility of 6.
But the graph does not show the representation of a balanced game.

According to Table 4 find the value of the game by means of the graphical method
applied to matrices 2 x n or m x 2.
R//:
For column the player: This means that if both players determine the probabilities of
the actions of each of their strategies, both will have the possibility of 2,5. This
represents a balanced game.

For the player row: This means that if both players determine the probabilities of the
actions of each of their strategies, both will have the possibility of 6.
But the graph does not show the representation of a balanced game.
 Theory will be
stimate the

Como P1 + P2 =1 P2= 1 - P1
Replace P2

3P1 +3 Strategy B
1P1 + 4 Strategy A

1P1 + 4
 1 - 1/2

E OF THE GAME

3 (1/2) + 3

Como Q1 + Q2 =1 Q2= 1 - Q1

-1Q1 + 6 Strategy 1
1Q1 + 3 Strategy 2
 1Q1 + 3

1 - 0/3

E OF THE GAME

-1(0/3)+6

l method

obabilities of
5. This

bilities of the
 Problem 5. Markov decision problem:

An insurance company charges its customers according to their accident history. If you have no
years are charged US $ 6000 (State 1); If you have had an accident in each of the last two yea
6300 (State 2). If you had accidents the first of the last two years US $ 5800 (State 3). The pro
according to historical data of three years are:

Table 1. Historical accident data

Table 5. Transition matrix of Markov chains

STATE E1 E2 E3
E1 0.18 0.42 0.4
E2 0.35 0.25 0.4
E3 0.2 0.15 0.65

Matriz de probabilidades de transición

X 0.18 0.42 0.40 =

Y P= 0.35 0.25 0.40 =
Z 0.20 0.15 0.65 =

La matriz de transicion para una cadena de Markov de n estados es una matriz de nxn con todo
de los registros de cada fila (0 columna) es igual a 1.

aX + bY + cZ + d = 0 donde,x,y,z son las variables y el termino independiente i

EC 1 0,18X-X + 0,35Y + 0,20Z =0

EC 2 0,42X + 0,25Y-Y + 0,15Z =0
EC 3 0,40X + 0,40Y + 0,65Z-Z =0
EC 4 X+Y+Z-1 =0

EC 1 _ 0,82X + 0,35Y + 0,20Z = 0

EC 2 0,42X -0,75Y + 0,15Z = 0
EC 3 0,40X + 0,40Y - 0,35Z = 0
EC 4 X+Y+Z-1 =0 Luego, la prima promedio

E0
W
0.2308

W
-0.82
0.42
0.40
1
According to Table 5 by applying the Markovian processes, ie finding the tra
equations of p * q, where p is the transition matrix and q the vector [W X Y
a.  What is the transition matrix resulting from proportionality according to
R//: (0,2308*6000+0,2359*6300+0,5333*5800)
b.  What is the average premium paid by a customer in Payoff, according to
R//: La prima promedio que deben pagar los clientes es 5,964
 history. If you have not had accidents the last two
each of the last two years you will be charged $
5800 (State 3). The probabilities of the state

1.00
1.00
1.00

∑ p*q=
1 EC 1 0,18X + 0,35Y + 0,20Z =X
1 q= ( X Y Z ) EC 2 0,42X + 0,25Y + 0,15Z =Y
1 EC 3 0,40X + 0,40Y + 0,65Z =Z
EC 4 X+Y+Z =1

matriz de nxn con todos los registros no negativos y con la propiedad adicional de que la suma

termino independiente igualadas a cero (0)

ego, la prima promedio que se paga en la campañia de seguros XYZ es:

E1 E2
X Y
0.2359 0.5333 1.0000

X Y Independiente Igual a:
0.35 0.20 0 -1.3878E-17
-0.75 0.15 0 -5.5511E-17
0.4 -0.35 0 2.7756E-17
1 1 -1 0.0000E+00

$ 5,964
ocesses, ie finding the transition matrix and solving the respective
and q the vector [W X Y Z]. Answer:
portionality according to the accident history?

er in Payoff, according to historical accident rate?

es es 5,964
 Problem 6. Markov decision problem:

Suppose you get 6 types of Jeans brands in the Colombian market: Brand 1, Brand
Brand 4, Brand 5 and Brand 6. The following table shows the odds that you continu
same brand or change it.

Table 6. Probabilities of change and permanence in the brand

STATE Brand 1 Brand 2 Brand 3 Brand 4 Brand 5 Brand 6
Brand 1 0.21 0.18 0.13 0.23 0.15 0.1
Brand 2 0.13 0.16 0.20 0.14 0.19 0.18
Brand 3 0.15 0.14 0.17 0.16 0.15 0.23
Brand 4 0.16 0.16 0.15 0.19 0.17 0.17
Brand 5 0.16 0.19 0.13 0.14 0.16 0.22
Brand 6 0.15 0.19 0.16 0.18 0.15 0.17

At present, brand, have the following percentages in market share respectively (19
17%, 15%, 19% y 12%) during week 4.

0.21 0.18 0.13

𝑋_4=𝑋_3 𝑃 0.13 0.16 0.20
0.15 0.14 0.17
(0,19 0,18 0,17 0,15 0,19 0,12) 0.16 0.16 0.15
0.16 0.19 0.13
0.15 0.19 0.16

P3 = 0.19 0.18 0.17 0.15 0.19

P4 = 0.1612 0.1697 0.1560 0.1728 0.1621

R// According the data of problm 5 the probability that each user stays with the ma
period for is:

P4=0,1612 0,1697 0,1560 0,1728 0,1621 0,1782

R// According the data of problm 5 the probability that each user stays with the ma
period for is:

P4=0,1612 0,1697 0,1560 0,1728 0,1621 0,1782

problem:

market: Brand 1, Brand 2, Brand 3,

e odds that you continue to use the

t share respectively (19%, 18%,

0.23 0.15 0.1

0.14 0.19 0.18
0.16 0.15 0.23
0.19 0.17 0.17
0.14 0.16 0.22
0.18 0.15 0.17

0.12 0.21 0.18 0.13 0.23 0.15 0.1

0.13 0.16 0.20 0.14 0.19 0.18
0.15 0.14 0.17 0.16 0.15 0.23
0.16 0.16 0.15 0.19 0.17 0.17
0.16 0.19 0.13 0.14 0.16 0.22
0.15 0.19 0.16 0.18 0.15 0.17

0.1782

t each user stays with the mark or change to another for

t each user stays with the mark or change to another for