DECISION SCIENCE

THE INNOVATIVE PROJECT

Submitted to:
Mr. Jagvinder Singh

Submitted By:
Misha(2K2O/UMBA/25)
Anshuman Singh(2K2O/UMBA/09)
Shreya Mukhija (2K2O/UMBA/40)
Shivam Shukla (2K2O/UMBA/39)

UNIVERSITY SCHOOL OF MANAGEMENT AND ENTREPRENEURSHIP

Delhi Technological University
DECEMBER 2020
 ACKNOWLEDGEMENT

We would like to express our humble thanks to our teacher and guide Mr. Jagvinder Singh
Sir, who gave us her valuable suggestions and ideas as and when needed.

We are sincerely thankful to him as he has been a constant source of advice, motivation and
inspiration. We are also grateful to him for giving him suggestions and encouragement
throughout the project work.

We are also grateful to our college for giving us the opportunity to learn and implement the
concepts and be able to analyze the results as per our understanding.

We would also like to extend our gratitude to our family and friends for constantly
motivating us to complete the project and providing us an environment that enhanced our
capabilities and knowledge.

2
 TABLE OF CONTENTS

S.No. Particulars

1. Objective of the Study

2. Introduction

3. Data

4. Model Formulation

5. Results

6. Conclusion

7. Reference
 OBJECTIVE OF THE STUDY

Selection of media and budget allocation is a major concern in advertising. It involves choosing the
appropriate media that is effective in reaching the target audience of the population in consideration.
Generally, predicting an optimal target audience exposure in selection of media is a complex problem. In
this study, a Linear Programming technique is used to investigate the budgetary allocation of a Company
for effective media selection planning. The problem is formulated using data from the company and other
media sources. The resulting linear programming model is analyzed using Quantitative Manager for
Business version and Linear Programming Solver version which embodies the simplex algorithm method.
The results show that the optimal target audience is 635,048,700. The optimum media mix which attracts a
significant audience exposure and generates the desired objective value for the advertising campaign
include; three (3) Television media outlet, two (2) print media outlet, and ten (10) radio media outlet.
 INTRODUCTION

Currently, the media selection planning has become increasingly complex which is evident in the
development of a variety of methods of solution which depends on the assumptions made in each case.
Thus, media selection planning has attracted a wide variety of model building than any other problem in the
marketing domain. For example advocates of the linear programming approach to the media selection and
planning problem would favor the following problem statement offered. “Media selection problem is to
allocate a scarce resource among a large number of alternatives so that the best possible contribution is
made to a central objective”. However the majority of approaches to the media problem suffer in that they
are static models apportioning the media budget to alternative media in order to maximize ‘effectiveness’ at
some point in time.

Media selection using a conceptual model is considered a serious challenge in Ghana. The fact is that most
companies are more inclined to very little or no conceptual approach in dealing with most media
campaigns. In some instances, it is considered as an insurmountable challenge in the country. It is in the
light of this that the linear programming conceptual model is adopted to expose the clear cut benefits of the
conceptual approach. The reason being that, conceptual models allow for optimization of the media
selection planning in such a way that, the effect of each variable present in relation to its cost and effect
which is completely absent in heuristic approach adopted over the years in Ghana is addressed.

In this paper, we devote attention to linear programming to examine the media selection planning problem
in Ghana.
 Theoretical Concept of the Simplex Algorithm

Linear Programming (LP) is the problem of maximizing or minimizing a linear function subject
to linear constraints.
A typical optimization problem is to find the best element from a given set (feasible set). In
Linear Programming (LP) problems, we need a criterion called an objective function
Z = f (x), and the feasible set, x is usually defined by
{x ∈ R |gi(x) ≤ 0, i = 1, . . . , m}.
Where gi(x) ≤ 0 are constraint functions.

Matrix Representation of Standard LP Problem

In the matrix notation, such an optimization problem can be formulated in the standard form as:

Maximize Z = ct x (2.7)
Subject to ax ≤ b (2.8)
x ≥ 0 (2.9)
Or in the canonical form as

Maximize Z = CT x (2.10)
Subject to ax = b (2.11)
x ≥ 0 (2.12)

where c = (c1, c2, . . . , cn)T , b = (b1, b2, . . . , bm)T , and x = (x1, x2, . . . , xn)T and A is m × n
coefficients matrix of rank m. The constraints of an LP problem may come with ≤, =, or ≥ which are
converted into equalities by adding slack and/or surplus variables.
 DATA

Data was collected on the media categories which the company intended to use for the advertisements.
These are as follows:

1. Television Media: GTV, TV3, Metro TV, Viasat 1, and Crystal TV

2. Print Media: Daily Graphic, Graphic Showbiz, Ghanaian Business, Junior Graphic, Ghanaian Sports,
the Mirror, Daily Guide, the Chronicle, Ghanaian Times, Ninety Minutes

3. Radio Media: Adom FM, Hitz FM, Joy FM, Happy FM, Peace FM, Asempa FM, Radio Gold, Citi FM,
Unique FM, BBC Radio, Choice FM, Hot FM, Obonu FM, Angel FM, Hello FM, Love FM, Kapital Radio,
Nhyira FM, Otec FM, Kessben FM, Radio Capital, Sunrise FM, Kyzz FM, Aseda FM, Melody FM, Volta
Star, Savana Radio, Uria Radio, Radio Upper West (UW), Radio Bar. Thus five TV media, ten Print media,
and thirty Radio media were used for the adverts.

Print Media Cost Per Advertisement Estimated Audience Exposure

Daily Graphic 3,928 3,928

Graphic Showbiz 970 970
Ghanaian Business 1, 614 1, 614
Junior Graphic 812 812
Ghanaian Sports 970 970
The Mirror 2,213 2,213
Daily Guide 2,909 2,909
The Chronicle 1,648 1,648
Ghanaian Times 1, 203 1, 203
Ninety Minutes 682 682
 MODEL FORMULATION

The Decision Variables

The decision variables consist of the number of times to use each medium in order to maximize the desired
target exposure whilst operating under the restrictions and policies of the company. Since all the media
outlets considered in this study fall into three major media categories, the decision

Variables are made up of five from TV, ten from print and thirty from radio media. The decision variables
are integer variables defined as:

Xti: Number of TV media (i = 1, 2 . . . 5) (t-index for TV)

Xpj: Number of Print media (j = 1, 2 . . . 10) (p-index for Print)
Xrk: Number of Radio media (k = 1, 2 . . . 30) (r-index for Radio)

Television media Cost (Gh¢) per Estimated Audience Exposure

advert

Meto TV 780 447,599

GTV 900 895,199
Viasat 1 500 413,169
TV3 427 516,461
Crystal TV 590 103,292

Radio media Cost (Gh¢) Estimated Media Cost (Gh¢) Est Audience Exp.
per advert Audience per advert
Exposure
Adom FM 103 65,655 Kapital 90 11, 197
Radio
Hitz FM 68 32, 827 Nhyira FM 25 78,381
Joy FM 151 87,541 Otec FM 42 55,986
Happy FM 25.9 32,827 Kessben FM 50 45,565
Asempa FM 42 21,885 Radio 25 25,888
Capital
Radio Gold 35.7 32,827 Sunrise FM 98 21,885
Citi FM 45 21,885 Kyzz FM 40 31,074
Unique FM 62 3,884 Aseda FM 35 3,305
BBC Radio 87 3,884 Melody FM 25 23,305
Choice FM 76.8 10,942 Volta Star 95 10,942
Hot FM 45 32,827 Savana 25 21,885
Radio
Obonu FM 48.7 10942 Radio UW 69 16,798
Hello FM 83.9 34,368 Radio Bar 45 98,651
Love FM 64 67,184 Peace FM 42 28,279
Angel FM 48.7 3,884 Uria Radio 26 20,597
The Objective Function
The objective of the problem is to allocate the advertising budget across the various media so as to
maximize the total target audience exposure for the campaign. Buying a unit space or time in each media is
expected to lead to a certain number of audience exposures. Therefore, audience exposure associated with
each media is a parameter of the objective function. Supposing that A = (Ati, Apj, Ark) and X = (Xti, Xpj,
Xrk) where Ati, Apj, Ark are respectively row vectors for TV, Print, and Radio audience exposures, Xti,
Xpj, Xrk representing the decision variables, then the objective function is formulated as follows;

Maximize Z = AX (1)
Where,

5 10 30

AX = ∑ AtiXti + ∑ ApjXpj + ∑ ArkXrk (2)

The Budget constraints

Let (Cti, Cpj, and Crk) be row vectors for TV, Print, and Radio cost of advert respectively. Let (Xti, Xpj,
Xrk) represent the decision variables of the problem whilst maintaining the definition of the other indices;
the total budget constraint is developed in a similar manner as in section (2) above.
• Total budget for TV media

• Total budget for Print media

• Total budget for Radio media

CtiXti ≤ 200, 000 (3)

i=1

10
Cpj Xpj ≤ 150, 000 (4)
j=1

30
CrkXrk ≤ 250, 000 (5)
i=1

Media Advert Constraints

• TV Media: Maximum advert for each TV media.
Xti ≤ 270 vi (6)
• Print Media: Maximum advert for each Print media.
Xpj ≤ {90, 12, 36, 24, 12, 12, 90, 12, 90, 36} vj (7)
• Radio Media: Maximum advert for each Radio media.
Xrk ≤ 270 vk (8)
The complete LP model resulting from equations 4.1 to 4.8, together with the non-negativity constraint has
45 decision variables and 48 constraints expressed as:
 RESULTS

The output of the model is presented in Table 4 which gives the recommended media mix for the study.
These include three Television media outlets, two print and ten Radio media outlets. These recommended
media mix identified generates an optimal value of 635,048,700 target audience exposure.

Variable and No. of adverts Variable and No. of adverts

Media Outlets Media Outlets

TELEVISION
Xt1: Metro TV 0 Xr24: BBC Radio 0
Xt2: GTV 94 Xr25: Choice FM 0
Xt3: Viasat 1 1 Xr26: Hot FM 270
Xt4: TV3 269 Xr27: Obonu 0
Xt5: Crystal TV 0 Xr28: Angel FM 0
Xr29: Hello FM 5
PRINT Xr30: Love FM 270
Xp6: Daily Graphic 0 Xr31: Kapital Radio 0
Xp7: Graphic Showbiz 0 Xr32: Nhyira FM 270
Xp8: Ghanaian Business 36 Xr33: Otec FM 270
Xp9: Junior Graphic 24 Xr34: Kessben FM 270
Xp10: Ghanaian Sports 12 Xr35: Radio Capital 270
Xp11: The Mirror 12 Xr36: Sunrise FM 0
Xp12: Daily Guide 0 Xr37: Kyzz FM 270
XP 13: The Chronicle 0 Xr38: Aseda FM 0
Xp14: Ghanaian Times 8 Xr39: Melody 270
Xp15: Ninety Minutes 36 Xr40: Volta Star 0
Xr41: Savana Radio 270
RADIO Xr42: Uria Radio 270
Xr16: Adom FM 270 Xr43: Radio UW 0
Xr17: Hitz FM 269 Xr44: Radio Bar 270
Xr18: Joy FM 270 Xr45: Peace FM 270
Xr19: Happy FM 270
Xr20: Asempa FM 270
Xr21: Radio Gold 270
Xr22: Citi FM 270
Xr23: Unique FM 0

The output shows that advertisement should be made 270 times in GTV, 199 times in Viasat 1 and 270
times in TV3 was enough to yield the overall exposure for the television media categories. For the print
(Newspapers), advertisement should be made 24 times in Junior Graphic and 36 times in Ninety Minutes as
indicated in Table 4.

For the radio stations, advertisements should be made 270 times in each of Happy FM, Radio Gold, Love
FM, Nhyira FM, Otec FM, Kessben FM, Radio Capital, Melody FM, Savanna Radio and Radio BAR, as
indicated by the basic variables in the solution given above. The variables such as Metro TV, Crystal TV,
Daily Graphic and Peace FM, having solution of zero at optimum level are non-economical to use. Though
estimated target exposure values in some of them may be encouraging such as Metro TV, Daily Graphic
and Citi FM, they are non-optimal decision variables.
 CONCLUSION

We have shown that out of the three major media categories which give a total of Forty-Five media outlets
in the country and advertising budget of Six Hundred Thousand Ghana Cedis (GH600,000), the optimal
target audience exposure is 552,597,527. The optimum media mix generating this objective value include;
three Television, two newspapers, and ten FM stations, which is achieved using our method. The sensitivity
analysis carried out shows that some media outlets such as Junior Graphic, and Ninety minutes, Radio Gold,
Happy FM & Love.

FM, Radio BAR etc., and GTV, Viasat 1 and TV3 had a tremendous impact on the optimal value since they
always remain in the basis regardless of any variation in the model.

Competing Interests- Authors have declared that no competing interests exist.

REFERENCE

 Cosgun O, Gultas I, Serarslan MN. Application of a mathematical model to an advertisement

research problem. An International Journal of Optimization and Control: Theories & Applications.
2013;3(1):23-33.
 Courtney Paulson, Lan Luo, Gareth M. James. Efficient large-scale internet media selection
optimization for online display advertising. Journal of Marketing Research. 2018; 55(4):489-506.

 Broadbent SR, Beale EML, Hughes PAB. A computer assessment of media schedules. Operational
Research Quarterly. 1966; 17:381-412.

 Kotler, Philip. Marketing decision making: A model building approach. New York, Galt, Rinehart
and Winston; 1971.

 Day RL. Linear programming in media selection. Journal of Advertising Research. 1962; 2(2):40-
44.

 Brown DB, Warshaw MR. Media selection by linear programming. Journal of Marketing Research.
1965; 2(1):83-88.

 Bass FM, Lonsdale RT. An exploration of linear programming in media selection. Journal of
Marketing Research. 1966; 3:79-88.

 Chang W, Ng M, Yuen W, Zhang S. A linear programming approach for determing optimal

advertising policy. IMA Journal of Management Mathematics. 2006; 17:83-96.

 Twum SB, Aspinwall E, Fliege J. A multicriteria optimization model for reliability design of series-
parallel systems-Part I. International Journal of Quality & Reliability Management; 2011.

 Murtagh BA, Saunders MA. Large-scale linearly constrained optimization. Mathematical

Programming. 1978; 14:41-72.