Operations Analytics (MBA3015)

Module 1

Selection of Operations
Strategy
 Decision Theory

Decision – Derived from Latin word ‘Decido’

Decision – is a settlement , a fixed intention

of bringing to a conclusive result , a
judgment and a resolution.

Decision – is a choice out of several options

Decision Making

Complex Concepts (T)

Simpler Concepts (B)

 DM – Decision Making is a Process

 It involves a series of conceptual stages starting

from elementary data and information to a
final conclusive report leading to management
action.
 Herbert Simon Model - I:D:C Model

 The Three Phases of the model

I - Intelligence Phase
D - Design Phase
C - Choice Phase
Block Diagram of the Model (HSM)

Intelligence

Design

Choice
 Important Steps in the various phases
of the model :

Problem Finding
Problem Formulation
Development of Alternatives
Proper Selection
 In programmed decision making, it is
necessary for the manager to enumerate all
the stage to the decision making situation,
and provide the necessary support through
rules and a formula for each one of them.

 In a closed decision making situation, the

programmed decision making system works
efficiently, while in the open situation it is
not efficient.
 Methods –
Statistical - conditional averages,etc
Optimization Techniques (. . . . ),
Payoff Analysis (. . . . ),
Decision Trees, etc

A decision tree can be adopted, if the decision

making situation can be described as a chain
of decisions.
Conditional Average( CA )
Ex-1 : Determine the conditional average row wise (CA-r) and column wise(Ca-c)
after excluding the extreme values from each row and each column. Consider a
random set of data generated in the range of 100 to 999 in 10x10 table. Use calculator
or Excel.

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10

R1 687 723 893 957 545 348 739 573 357 525

R2 727 230 931 348 675 649 171 108 283 923

R3 657 795 731 733 162 778 600 324 167 951

R4 839 136 603 794 588 396 524 155 594 130

R5 127 166 569 555 445 547 886 534 742 399

R6 308 860 101 798 574 234 958 206 854 389

R7 429 891 304 780 961 235 975 743 132 278

R8 541 163 920 576 697 135 954 314 171 696

R9 718 684 468 218 950 671 186 724 355 116

R10 797 252 889 517 417 846 267 141 913 312
Solution : Conditional average is calculated by excluding the highest and lowest
value in a row or a column.
Ex : in row 1 :
CA-r = [sum (values in row 1) – max (value) – min (value)] /8
= (( 687+723+ …… +525)-957- 348 )/8
= (6347 – 957 – 348 )/8
= 630.25
Similarly determine for other rows and for columns. This is useful in de-seasoning of
data in time series analysis.
Generate the results in Excel also. Hint : use functions : SUM(), Max(), Min().
Payoff analysis : Determine Ʃp.x
Ex-2 : Determine the value of weighted payoff Ʃp.x (also known as EMV in payoff
analysis) for each column. Consider a random set of data generated in the range of
100 to 999 in 10x10 table. Use calculator or Excel. Consider weights as indicated in
the table. Determine highest and lowest EMV.

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10 Wt
R1 687 723 893 957 545 348 739 573 357 525 5%
R2 727 230 931 348 675 649 171 108 283 923 10%
R3 657 795 731 733 162 778 600 324 167 951 6%
R4 839 136 603 794 588 396 524 155 594 130 4%
R5 127 166 569 555 445 547 886 534 742 399 20%
R6 308 860 101 798 574 234 958 206 854 389 15%
R7 429 891 304 780 961 235 975 743 132 278 8%
R8 541 163 920 576 697 135 954 314 171 696 12%
R9 718 684 468 218 950 671 186 724 355 116 7%
R10 797 252 889 517 417 846 267 141 913 312 13%
Solution : EMV is calculated by wt and corresponding payoff in each column and
adding the products.
Ex : For Column 1 :
Ʃp.x = [(5% x 687) + (10% x 727 )+ ……. + (13% x 797) ]
= [(34.35) + (72.7)+ ……. + (103.61) ]
= 504.74
Similarly determine for other columns.
Generate the results in Excel also. Hint : use functions : SUMPRODUCT(),
max() , min(). Results are shown in the table below.
 The Choice Phase

CERTAINTY Si

RISK P (e)
i

UNCERTAINTY M (Si) P (e)

i
 The Selection (Choice) Phase

MAXIMIN, MINIMAX, MAXIMAX

UNCERTAINTY LAPLACE, H-α , REGRET, etc.

PAYOFF- EMV, EOL, EVPI ,

RISK DECISION TREE, UTILITY etc.
 The payoff Analysis – When all the alternatives and
their outcomes are not known with certainty, the
decision is made with the help of payoff analysis.

 Components:
 Rows : alternatives
 Columns : conditions/ states of the nature
with the probability of occurrence.
Caselet :
A Herbal care products company is faced with a
challenge of launching a new product in
competition to another similar product in the
market. There is an element of uncertainty as to
whether a particular product launched would be
well accepted or not ? A market survey was
conducted and data was collected pertaining to
possible sales level, profit and loss. What would be
the optimum decision ?
 Information Table
Sales Level – product type

PROFIT : LOSS (Rs. Lakh)

Estimated level of sales (units)
25000 20000 15000
LAUNCH
30 : 20 20 : 15 15 : 10
PRD-1
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3
 DM under Uncertainty - Maximin
PROFIT : LOSS (Rs. Lakh)
Estimated level of sales (units)
25000 20000 15000
LAUNCH
30 : 20 20 : 15 15 : 10
PRD-1
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3

(approach - p) Min profit (15,20,5)

MAXIMIN CRITERION : LAUNCH PRD-2
 DM under Uncertainty-Minimax

PROFIT : LOSS (Rs. Lakh)

Estimated level of sales (units)
25000 20000 15000
LAUNCH
30 : 20 20 : 15 15 : 10
PRD-1
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3
(approach - p) Max loss (20,40,60)
MINIMAX CRITERION : LAUNCH PRD-1
 DM under Uncertainty - Maximax
PROFIT : LOSS (Rs. Lakh)
Estimated level of sales (units)
25000 20000 15000
LAUNCH
30 : 20 20 : 15 15 : 10
PRD-1
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3

(approach - o) Max profit (30,50,70)

MAXIMAX CRITERION : LAUNCH PRD-3
 DM under Uncertainty – H-alpha
PROFIT : LOSS (Rs. Lakh)
Estimated level of sales (units)
25000 20000 15000
LAUNCH
30 : 20 20 : 15 15 : 10
PRD-1
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3

(approach – o & p) H-alpha CRITERION

 DM under Uncertainty
PROFIT : LOSS (Rs. Lakh)

Estimated level of sales (units)

H- α Criterion
25000 20000 15000
LAUNCH
Decision Index
30 : 20 20 : 15 15 : 10
PRD-1
DI = (α)M +(1- α)m
LAUNCH
40 : 35 50 : 40 20 : 15
PRD-2
LAUNCH
70 : 60 30 : 25 5:3
PRD-3

Assume α = 60%
D1 = 0.6(30) +(1-0.6)15 = 24
D2 = 0.6(50) +(1-0.6)20 = 38
D3 = 0.6(70) +(1-0.6)5 = 44 (max)

LAUNCH PRD-3
 Example : DM under Risk - Pricing problem- EMV

C-1 C-2 C-3

Strategy
Pi() 0.5 0.2 0.3
No change in
price
4 5 8

Increase the
price
6 4 3

Decrease the
price
10 12 4

What would be the optimum strategy under the stated

conditions ?
 Example : DM under Risk - Pricing problem- EMV

C-1 C-2 C-3 EMV

Strategy
Pi() 0.5 0.2 0.3
No change in
4 5 8 5.4
price
Increase the 4.7
6 4 3
price
Decrease the 8.6
10 12 4
price

Decision : Decrease the price

Payoff analysis : EMV= Ʃp.x
Ex-6 : A company has 11 strategies of which one has to be selected. Determine the
value of EMV for each strategy of the company. Different strategies are indicated in
the table below with respective payoff and probabilities under different conditions C1
to C6. Identify the most suitable strategy for the company.

C-1 C-2 C-3 C-4 C-5 C-6

P() 0.2 0.15 0.12 0.18 0.15 0.2

No change in price 4 5 5 6 8 8

Increase the price 10% 6 4 3 6 2 3

Decrease the price 10% 10 12 8 10 9 4

Increase the price 15% 5 6 8 9 6 5

Decrease the price 15% 8 24 5 26 2 4

Increase the price 20% 13 11 9 11 6 11

Decrease the price 20% 9 17 11 1 14 10

Increase the price 25% 8 21 1 14 17 12

Decrease the price 25% 12 15 1 4 21 16

Increase the price 30% 8 12 8 12 8 9

Decrease the price 30% 11 9 8 10 9 5

Solution : EMV is calculated by wt and corresponding payoff in each row and adding
the products.
Ex : For strategy row 1 :
Ʃp.x = [(0.2x 4) + (0.15 x 5 )+ ……. + (0.2 x 8) ]
= 6.03
Similarly determine EMV for other strategies rowwise.
Generate the results in Excel also. Hint : use functions : SUMPRODUCT(),
max(), VLOOKUP() .Results are shown in the table below.
 The concept of utility relates to the money value
considered by the decision maker.

 The monetary values of the outcomes are replaced

by the utile values.

 Decision tree analysis – when a decision maker

must make a sequence of decision, the decision
tree analysis is useful in selecting the set of the
decisions.
DT Caselet– Borewell Problem
A bore-well company wants to decide upon
drilling a well for a farm in a village. In
that village, only 40% of the wells drilled
were successful at 200 ft depth. Some of
the cases where they did not get water at
200 ft, drilled further up to 250 ft but
only 20% of the bores yielded water at
250 ft. Cost of drilling is Rs.50 per foot. It
is estimated that Rs.18,000 would be paid
during a 5 year period in the present
value terms, if the water is bought from
the neighbor rather than go for the well
which would have a life of 5 years.
Identify the options ?
There are three options :

(a) The well be drilled up to 200 ft

(b) If no water is found at 200 ft, the well be drilled up to

250 ft?

(c) Water be bought from the neighbor?

Decision tree
(Scenario 1 : @200’ succ=40%, @250’ succ=20%)
k
t r uc
s
t er
Wa 0.4
k
t r uc
A No e rs
00’
t
2 W Wa 0. 2
upto at
er ’ B No
ill 50 W
Dr st
ru pt o 2
0.
at
er

0.
u 8

6
ck il l st
Dr ru
ck
1
Do 2
no Do
td no
ri l td
l ril
l
Decision tree
(Scenario 1 : @200’ succ=40%, @250’ succ=20%)
EO
k
20140 t r uc Rs.12,500
s Rs.10,000
t er EO
Wa 26900 uc
k
0. 4 rs
t r
0’ A No t e
EO 20 W Wa 0. 2
upto at
er ’ B No
18000 ill 50 W
Dr st
pt o 2
0.
at
er

0.
ru u 8

6
ck il l st
Dr ru
ck
1 Do
no 2
td Do Rs.
ril
l
EO no (12,500+18,000)=Rs
td .30,500
26900 ril
l
Rs.18,000
Rs.
(10,000+18,000)=Rs.
28,000

Min EO – is Rs 18000, therefore “ Do not Drill "

Decision tree
(Scenario 2 : @200’ succ=80%, @250’ succ=90%)
k
t r uc
r s
te
Wa
k
t r uc
A No te rs
200’ W W a
upto at
er ’ B No
ill 50 W
Dr st
ru pt o 2 at
er
ck i ll u st
ru
Dr ck
1
Do 2
no Do
td no
ri l td
l ril
l
Decision tree
(Scenario 2 : @200’ succ=80%, @250’ succ=90%)
EO
k
10860 t r uc Rs.12,500
r s Rs.10,000 EO
te
Wa 14300 uc
k
0. 8 rs
t r
A e
00’
No t
EO 2 W Wa 0. 9
p to at No
10860 ill u er 50
’ B W
Dr st
pt
o 2
0.
at
er

0.
ru u 1

2
ck i ll st
Dr ru
ck
1 Do
no 2
td Do Rs.
ril (12,500+18,000)=Rs
l
EO no
td .30,500
14300 ril
l
Rs.18,000

Rs.
(10,000+18,000)=Rs.
28,000

EO is Rs 10860 (least) at node-1, therefore “ Drill up to 200’ “

EO is Rs 14300 (least) at node-2, therefore “ Drill up to 250’ “
Problems on Risk - EMV
Sunrise Resorts wants to decide upon the best
alternative amongst the four locations for which the
annual revenues are estimated along with their
respective probability. The estimates were based upon
the survey conducted by experts at the location.
Assume initial investment and roi same for each
alternative.
Mahabaleshwar Munnar Kodaikanal Darjeeling
Rs-L P() Rs-L P() Rs-L P() Rs-L P()
85 .15 75 .3 80 .2 65 .6
75 .3 80 .4 60 .4 70 .2
90 .15 60 .1 70 .2 75 .1
60 .4 65 .2 75 .2 80 .1
Mahabaleshwar Munnar Kodaikanal Darjeeling
Rs-L P() Rs-L P() Rs-L P() Rs-L P()
85 .15 75 .3 80 .2 65 .6
75 .3 80 .4 60 .4 70 .2
90 .15 60 .1 70 .2 75 .1
60 .4 65 .2 75 .2 80 .1

EMV computation (Rs Lakh)

72.75 73.50 69.00 68.50

Problems on Risk
Gokul Bakery wants to decide upon the stocks of
special milk fruit bread on Mon &Thr. The profit on a
loaf of bread is Rs 4.0 and the loss for not selling the
loaf is Rs 8.0 at the end of the third day. A market
survey was conducted to determine the demand of
bread on these days for 1 yr. The data is indicated in
the(a)table.
Monday (b) Thursday
Demand N Demand N
5000 18 5000 16
6000 18 6000 18
7000 16 7000 18
Monday

Demand N Probability

5000 18 0.35

6000 18 0.35

7000 16 0.30
EMV Approach -
Obtain profit matrix based on demand and possible stocks

Stock (Possible Action)

Pi Demand 5000 6000 7000

0.35 5000
0.35 6000
0.30 7000
Obtain profit matrix based on demand and possible stocks

Stock (Possible Action)

Pi Demand 5000 6000 7000
0.35 5000 20000 12000 4000
0.35 6000 20000 24000 16000
0.30 7000 20000 24000 28000

EMV 20000 19800 15400

EOL Approach -
Obtain opportunity loss matrix based on demand and possible stocks

Stock (Possible Action)

Pi Demand 5000 6000 7000

0.35 5000
0.35 6000
0.30 7000
EOL Approach -
Obtain opportunity loss/loss matrix based on demand and possible stocks

Stock (Possible Action)

Pi Demand 5000 6000 7000

0.35 5000 0 8000 16000

0.35 6000 4000 0 8000
0.30 7000 8000 4000 0

EOL 3800 4000 8400