COURSE OUTLINE

COURSE NAME: Operations Research

COURSE CODE:

Programme MBA
Academic Session 2022-24
Semester II
Credits 3
No of Sessions 45
Prerequisite Mathematics at high school
level and basic statistics

MBA Program Educational Objectives: (PEOs)

PEO1: Graduate will grow personally and professionally to be successful in dynamic
business environment globally.
PEO2: Graduate will lead with creative, analytical and entrepreneurial skills to help
organizations achieve a competitive advantage.
PEO3: Graduate will contribute as socially and ethically responsible citizen or the
sustainable development of environment and society at large.

MBA Programme Outcomes (POs)

PO1: Apply knowledge of management theories and practices to solve business problems.
PO2: Foster analytical and critical thinking abilities for data-based decision making.
PO3: Ability to develop Value based Leadership.
PO4: Ability to understand, analyze and communicate global, economic, legal, and ethical
aspects of business.
PO5: Ability to lead themselves and others in the achievement of organizational goals,
contributing effectively to a team environment.
PO6: Ability to conduct investigation of complex problems using modern tools and
techniques.
PO7: Apply the various forms of business communication, supported by effective use of
appropriate technology, logical reasoning, articulation of ideas.
PO8: Apply innovative ideas and knowledge of business project management principles as
an entrepreneur/team member/ team leader to develop and manage projects in
multidisciplinary environments.

Course Overview
Operations Research, an approach to decision making based on the scientific method, makes
extensive use of quantitative analysis. This course deals with both deterministic and
stochastic problems, and they will be analyzed based on following models and methods:
Linear programming, network models, game theory, and simple queuing theory.Students will
be able to perform sensitivity analysis and describe the economic information that can be
drawn from the analysis. They will be able to create different models based on verbal
descriptions and discuss the pros and cons of these models alongwith the associated solution
methods in view of specific problem.
Course Objectives
The course aims to acquaint students with the application of Operations Research in
industry and business that involves limited resources and need to find the optimal solution
for decision making.

Course Outcomes
After completing the course, students would be able to:
CO1 Make use of LPP technique to formulate a given problem in mathematical language
and obtained the optimal solution and interpret the decision. (K3)
CO2 Apply the concepts of transportation and assignment problem to find the optimum
allocation in order to either minimize total cost or maximize profit. (K3)
CO3 Apply and analyse the concepts related to decision making under uncertainty and risk
for making more informed decision. (K4)
CO4 Determine a stable strategy to optimize the Payoff in different conditions of business
environment by utilizing the concept of game theory. (K5)
CO5 Evaluate the queue length, system length and optimize the waiting time of customer
by applying the concepts of queuing theory. (K5)

CO-PO matrix(3/2/1 indicates strength of correlation where 3- Strong, 2- Medium, 1-

Low)

CO/PO PO1 PO2 PO3 PO4 PO5 PO6 PO7 PO8

CO1 1 3 2 2 3 2

CO2 2 3 2 2 1

CO3 1 3 2 3 1 2 1

CO4 2 3 1 3 3 1

CO5 1 3 2 2 1

Unit I: IntroductionLectures: 10
Introduction to OR, Use, scope and applications of OR, Introduction to Linear Programming
problem, Mathematical form of LPP, Solution to LPP, Graphical Method and its limitations,
Types of solution in graphical method, Multiple solution, unbounded solution and infeasible
solutions, Solution of LPP by simplex method, Maximization and Minimization examples,
Duality, Sensitivity Analysis using Excel Solver
Case: The Locomotive Fueling Problem (LFP) (Chapter2)
Link:https://eco.nahrainuniv.edu.iq/wp-content/uploads/2020/12/
Katta_G._Murty_eds._Case_Studies_in_OperationsBook4You.pdf

Unit II: Transportation and Assignment Problems Lectures: 8

Introduction to Transportation problem, Mathematical form of TP, North West Corner
Method, Vogel’s approximation method, least cost Method, MODI method for optimum
solution, MODI method using excel solver, Introduction to Assignment Problem,
Mathematical form, Method of solution- Hungarian method, unbalanced assignment problem
(Excel Solver), restricted assignment problem, Applications of assignment problem in
business
Case: Optimization_of_Personnel_Assignment_Problem_Based
Link:doi:10.1088/1757-899X/300/1/012005

Unit III: Decision Theory Lectures: 10

Decision Theory (Decision Under certainty, risk and Uncertainty, Marginal Analysis,
Decision tree Analysis), Maximax, Minimin, Maximin, Minimax, Laplace, Hurwicz and
Minimax Regret Criteria. EMV Criteria, Decision Tree Analysis and Tree Plan with solver.
Case:Decision Tree Cart-Retail Case
Link: http://ucanalytics.com/blogs/decision-tree-cart-retail-case-example-part-5/

Unit IV: Game Theory Lectures: 9

Introduction of game theory, concepts, Two persons game, Zero sum Game, Strategy, Mixed
Strategy game & Pure strategy games, Methods of solution, Maximin-Minimax principle,
Saddle Point and Saddle Point Method, Probability Method, Graphical Method, Principle of
Dominance.
Case: Product price control using game theory
Link: https://doi.org/10.1063/1.4887749

Unit V: Queuing Theory Lectures: 8

Introduction to queuing theory, concept of arrival pattern, Service Pattern, Queuing
Discipline, Physical capacity of system, No of servers and multi- stage service, Application
of Queuing Model: M/M/1/∞
Case: Restaurant Model
Link:https://www.researchgate.net/publication/
314273787_Case_Study_for_Shuruchi_Restaurant_Queuing_Model?enrichId=rgreq-
55153fdbc3af4d05815a931e04c6dbb6-
XXX&enrichSource=Y292ZXJQYWdlOzMxNDI3Mzc4NztBUzo2NDU5OTQwOTgyMjEw
NTZAMTUzMTAyODQwMzA1Ng%3D%3D&el=1_x_2&_esc=publicationCoverPdf

Text Books
1. J.K Sharma, Operations Research (Theory & Applications), 6th edition, Trinity Press

Reference Books
1. N.D Vohra, Quantitative Techniques in Management, 2nd Edition, Tata Mcgraw Hill
2. Hamdy A Taha,Operations Research-An Introduction 7th Edition, Prentice Hall
3. Peter C Bell- Management Science/ Operations Research: A strategic perspective,
South-Western College Pub

FULL CASES to be pasted here

Case: The Locomotive Fueling Problem (LFP) (Unit 1)
The problem deals withN[= 214] trains hauled by L [= 214] locomotives on a railroad network consisting of Y [=
73] yards over a 2 week planning horizon. The yard to yard distances, over the railroad network is given; the
average yard to yard distance is 285.66 miles with standard deviation of 44.54 miles, median and mode of 300
miles. All locomotives are assumed to be identical in performance.
Each train visits a sequence of yards (referred to as route in the paper). For example, the route for the train T10
is the sequence (Y43, Y16, Y11, Y2, Y3, Y29,Y28, Y23) of yards, where Y43, Y23 are the origin, destination
yard; and Y16,Y11, Y2, Y3, Y29, Y28 are all intermediate yards in that order in this route. A few characteristics
of the trains included in the case are:
• All the 214 trains operate daily. Thus 214 trains originate every day from respective originating yards.
• 52 trains reach the destination yard the same day it leaves the originating yard. The remaining 162 trains reach
the destination yard the next day.
• 135 trains ply between two yards only. Of these 135 trains, only 49 trains reach destination the same day
(examples are trains T2 and T4); the remaining 86 trains reach destination the next day(examples are trains T1
and T3). 35 trains traverse only one intermediate yard between origin and destination yards.
Of these 35 trains, 11 trains reach the intermediate and destination yards the next day (examples are trains T35
and T51) and 21 trains reach the intermediate yard on the starting day, but reach the destination yard the next
day(examples are trains T36 and T52).
• 20 trains traverse two yards between origin and destination yards. Of these 20 trains, 14 trains reach the first
intermediate yard the next day (examples are trains T13 and T14), 18 trains reach the second intermediate yard
the next day (examples are trains T13, T14, T16, and T94) and all trains reach their destination the next
day.
• 16 trains traverse three yards between origin and destination yards. Of these 16 trains, 7 trains reach the first
intermediate yard the next day (examples are trains T7 and T8), 10 trains reach the second intermediate yard the
next day (examples are trains T7, T8, T33, and T40), 15 trains reach the third intermediate yard the
next day (examples are trains T7, T8, T33, T34, and T39) and all trains reach their destination the next day.
Each route may be hauled by a different locomotive on different days; the allocation of locomotives to routes is
given as data. For hauling each train in this case study, only one locomotive is used. Each locomotive may haul
different trains on different days.
All the routes operated in the case study problem can be grouped into a set of pairs, each pair operating between
a pair of yards in the forward and reverse directions; but the set of yards visited in the two directions for a route-
pair may be different. Every route pair has a dedicated pair of locomotives operating it. When we refer to
ayard on a route, we mean either the origin or destination yards of the route, or an intermediate yard where the
train has a scheduled stop. An example is locomotives L1, L2 hauling trains T1 on route (Y25, Y19) and T2 on
route (Y19, Y25); with L1, L2 hauling T1, T2 respectively on days 1, 3, 5, 7, 9, 11, 13; and L2, L1 hauling
T2, T1 respectively on days 2, 4, 6, 8,10, 12, 14 of the planning horizon. Another example is locomotives L5,
L6 hauling trains T5 on route (Y36, Y60, Y62), T6 on route (Y62, Y36) on alternate days.
The locomotive of each train can be refueled by fueling trucks positioned at any of the yards on its route except
the destination yard, but the total number of refuelings on any route should be ≤2. All locomotives have the
same fuel capacity of 4500 gallons; and the fuel consumption (3.5 gallons of fuel per mile) on any route is
independent of the route and the locomotive operating the train. Fuel consumption for a locomotive traveling
between any two yards can be determined using the given table of inter-yard distances. Fuel cost at different
yards varies between $ 2.90 to 3.56/gallon (with average of $ 3.13 and standard deviation of $ 0.17). Every time
a locomotive is refueled, a setup cost of $ 250 has to be paid in addition to the priceof the fuel loaded. Fuel is
dispensed by fueling trucks positioned at yards, each having a maximum capacity of 25,000 gallons/day and
involving a one-time contracting cost of $ 8000 for the 2 week planning horizon (the contracting cost is $ 4000
per week per truck). The problem statement allows each locomotive to start on the very first trip with any
feasible amount of fuel (referred to as “initialfuel” in this paper) without any cost incurred; the locomotive
should be left with the same amount of fuel after completing the last trip in the planning horizon.
Case: Decision Tree Cart-Retail Case (Unit 3)
You have recently joined in as the chief analytics officer & business strategy head at an online
shopping store called DresSMart Inc. that specializes in apparel and clothing. One day you had the
chief marketing officer of the company come rushing to you office looking unusually worried . The
board of directors has given him hefty targets for sales and slashed his marketing budgets into half at
the same time. You immediately identify that you are dealing with a common business problem of
improving business revenue with reduced cost. You have also realized that this is a great opportunity
for you to establish analytics practices in the company since there is a quick opportunity for you to
improve the P&L (income statement).

Additionally, the CMO informed you that last year they had carried out marketing campaigns with
different offers on the product catalog. A direct mailing product catalog was sent to some hundred
thousand customers from the base of over a couple of million customers last year with the response
rate of 4.2%. The direct mailers were later followed up with SMS and email messaging.

To explain your strategy to the CMO, you drew a quick and dirty campaign P&L statement on the
white board in your office. The following is a version of your drawing

Retail Case – Decision Tree (CART)

Back to our retail case study example, where you are the Chief Analytics Officer & Business Strategy
Head at an online shopping store called DresSMart Inc. that specializes in apparel and clothing. In this
case example, your effort is to improve a future campaign’s performance. To meet this objective, you
are analyzing data from an earlier campaign where direct mailing product catalogs were sent to
hundred thousand customers from the complete customer base of over a couple of million customers.
The overall response rate for this campaign was 4.2%.

You have divided the total hundred thousand solicited customers into three categories based on their
past 3 months activities before the campaign. The following is the distribution of the same. Here, the
success rate is the percentage of customers responded (r) to the campaigns out of total solicited
customers.

Activity in Number Campaign Results Success Ra

the of te
Last Quart Solicited Responde Not Responde
 d (r) d (nr)
er Customers
low 40000 720 39280 1.8%
medium 30000 1380 28620 4.6%
high 30000 2100 27900 7.0%

As you know, CART decision tree algorithm splits the root node into just two child nodes. Hence for
this data, CART can form three combinations of binary trees as shown in the table below. We need to
figure out which is the best split among these 3 combinations. The results for the same are shown in
the table below.

Ψ(L Ψ(Pi
arge ck Good
P(k P(k Piec Cher
Left Right P P ness
|L) |R) e) ries)
Node Node L R of
=a =b
2PLP Σ(a- Split
R b)
r: r:
0.0 0.0
0 0 18 58
Medium 0.038
Low . . 0.48 0.080
+High nr: nr: 4
4 6
0.9 0.9
82 42
r: r:
0.0 0.0
0 0 30 70
Low+M 0.033
High . . 0.42 0.080
edium nr: nr: 6
7 3
0.9 0.9
70 30
r: r:
0.0 0.0
0 0 40 46
Low+hi 0.004
Medium . . 0.42 0.011
gh nr: nr: 8
7 3
0.9 0.9
60 54

Let me help you out with the calculation of each column for the above tree. We will use the first row
(i.e left node: Low and right node: Medium+High) for the following calculations and then you could
do the rest of the calculations yourself. To start with we have calculated P L and PR in the following
way:
Now the calculation for Ψ(Large Piece) is simple as shown below:

Now, let’s come to the second part of the equation that is Ψ(Pick Cherries). Remember, r represents
responded and nr represents not-responded customers for our campaign’s example.

You may want to calculate the other two terms (i.e r: P(k|R), and nr: P(k|R)) yourself before plugging
them in the following equation to get the value for Ψ(Pick Cherries).

This leaves us with one last calculation for the last column i.e. goodness of split which is:

The final task now is to find the maximum value for goodness of split in the last column. This will
produce the following decision tree through the CART algorithm with Low on the left node, and
Medium+High on the right node.
Decision Tree – The CART Algorithm Final Result
This is an important business insight as well that people with higher activity tend to respond better to
campaigns. I agree it was clear from the first table at the top as well, but we have learned the science
of creating decision tree using the CART algorithm in the process. This is extremely useful when you
are dealing with a large dataset and want to create decision tree through recursive partitioning.