International Journal of Mathematics and Statistics Invention (IJMSI)

E-ISSN: 2321- 4767 P-ISSN: 2321-4759

www.ijmsi.org Volume 8 Issue 6 || June, 2020 || PP 14-23

Application of Markov Analysis to Consumers Preference

T. E. Efor and Chukwu John
1,2
Department of Industrial Mathematics and Applied Statistics, Ebonyi State University, Abakaliki, Nigeria
Corresponding Author: T. E. Efor

ABSTRACT
In real world today, the changes in consumers preferences is a basis for competitive advantage. This paper is an
attempt to understand these changes in consumer’s preference. The work investigates the changes in preference
of a population after a period of time. The state of equilibrium of the market was determined. Transition
Matrices are used to perform a Markov analysis on the preferred transport services of students in Ebonyi State
University. A survey was conducted on 54 undergraduate Students of the school in particular, students of the
Department of Industrial Mathematics and Applied Statistics. The data extracted from this survey were cast into
a Markov Matrix and was analyzed. The result was used to forecast the preferences of consumers in the market
environment in the future. Also the result showed that consumers exhibit different preferences with respect to
some attribute such as; comfort, speed, availability and price.
KEYWORDS: Brand loyalty, Consumer preference, Markov chain, Market equilibrium, Steady State
----------------------------------------------------------------------------------------------------------------------------- ---------
Date of Submission: 13-06-2020 Date of Acceptance: 29-06-2020
--------------------------------------------------------------------------------------------------------------------------------------

I. INTRODUCTION
Markov Analysis is a stochastic process that is used to examine randomly changing system. It is
probabilistic information about a decision situation that can help the decision markers in making decisions. It is
used to forecast the variable whose future is variable to its current state. In market today building a competitive
advantage takes a brand/company that understands the preference of the consumers that patronize that market
and also the understanding of why consumers become loyal or disloyal to a brand. Markov analysis helps to
understand the movement of consumers from one brand to another in a market. In 1970, the Russian
Mathematician and Author Andreyevich introduced the basic concept of Markov Chain method. Since that time
many authors has been working on Markov chain and have helped it to develop. Researchers have used Markov
chain method intensively for research conducted on social topics as the brand selection of customers, income
distribution, occupational mobility and immigration as a geographical structure.
In marketing, Markov Analysis has been widely studied by many researchers such as, [1] applied it in
brand switching studies and market share forecasting and observed that Markov brand switching model studies
customer loyalty and forecasts the brands, Products, or services that a customer is likely to purchase next, [2]
tries to model Brand switching in consumers products with Markov analysis and transition matrix, using
different brands of soft drinks, [3] studied the Application of Markovian models and transition Matrix to
analyze the workforce movement in Jordanian productivity company‟s statistics. Michael i n [ 4 ] studied the
applications of finite Markov chain models to management. In [5], Aparna and Sarat researched on analysis
of brand loyalty using homogeneous Markov models.
In this study we applied Markov chain analysis to understand the uncertainties of student‟s preference
of transport services, investigate the changes in preference of a population after a period of time and determined
the state of equilibrium of the market. Here we considered three transport services; Bus, Bike and tricycle
(Keke Napep) a n d t r i e d t o f i n d o u t reasons for discontinuity of students preference for transport
services.

II. SOME DEFINITIONS

Definition 1.1 Markov chain is a mathematical and scientific method used to analyze the behavior of
different phenomena (system/state) during the current period in order to predict the behavior of these
phenomena‟s in the future in any later periods according to Touama (2015). It can also be defined Godwin et al
(2013), as a dynamic quantitative model with markov property; the probability of moving to a state (X) at
time (n+1) depends only on current state (immediate past purchase behavior) and not on the sequence of
the entire previous states.

Definition 1.2 Brand loyalty refers to consumer‟s behavior of repeatedly purchasing/patronizing a specific
brand over a certain period of time. According to [2], brand loyalty is consumer behavior of choosing to stick

DOI: 10.35629/4767-08061423 www.ijmsi.org 14 | Page

 Application of Markov Analysis to Consumers Preference

with a particular brand, despite the availability of other closely related and similar brands in the market.
Definition 1.3
Brand Switching is choosing to switch from routine use of one product or brand to steady use of different
but similar product. When consumers switch from one brand to another, building a picture of likely
brand switching behavior occurs. It is also the behavior of consumers to rotate between use of different, but
similar products over a period of time.

III. TRANSITION/STOCHASTIC MATRIX

Definition 1.4 A transition matrix is a square matrix whose entries are probability vectors, which is used to
describe the transitions of a Markov chain. It is also called a probability matrix or Markov matrix.
There are several types of transition matrix namely:
Right Transition Matrix: A real squared matrix with each row summing to 1.
Left Transition Matrix: A real squared matrix with each column summing to 1
Double Stochastic Matrix: A squared matrix of non negative real numbers with each row and column
summing to 1.
Because of the transition (changes) between different states, conditions or choices, markov analysis deals with
transition probabilities. Transition matrix is used to exhibit/model these transition probabilities.

IV. ASSUMPTIONS:
In Markov Analysis for a Process (stochastic process) to be called a markov process, it must be
characterized by some assumptions: An analysis of the markov method is based on the fundamental assumption
that: any system dealt with in the first instance is in its initial state, in preparation for the transition to another
state. This assumption is based on a certain probability law called the (Transition Probability), which are known
as “transition Probabilities of a particular case to another case during a certain period of time”.
For example, the probability of transition phenomenon of the case (i) in the current period (n) to another
state (j) in the later period (n+1) Writes as follows according to Touama, (2015)

𝑃 𝑋𝑛+1 = 𝑗/𝑋𝑛 = 𝑖 = 𝑃𝑖𝑗 ∀𝑖, 𝑗

Where;
Xn: Value of the phenomenon (state) in the current period (n),
Xn+1: Value of the Phenomenon (state) in the subsequent period (n+1).
𝑃𝑖𝑗 is the probability of transmission from the phenomenon of state (𝑖) to the state (𝑗).

Amare et al in [1] made the f o l l o w i n g a s s u m p t i o n s before formulating a markov chain model:

- There is finite number of possible states.
- States are both collectively exhaustive and mutually exclusive.

- The transition probabilities depend only on the current state of the system, that is, if the current
state is known, the conditional probability of the next state is independent of the states prior to the present
state.

- The long-run probability of being in a particular state will be constant over time.

- The transition probabilities of moving to alternative states in the next time period, given a state in the
current time period must sum to 1.

V. CLASSIFICATION OF THE STATES IN A MARKOV CHAIN

According to [7]; The states of a markov chain can be classified based on the transition probability (𝑃𝑖𝑗 ) of P;
ABSORBING: A state j is absorbing if it returns to itself with certainty in one transition. That is 𝑃𝑖𝑗 = 1.

TRANSIENT: A state j is transient if it can reach another state but cannot itself be reached back from another
state. That is;

lim𝑛→∞ 𝑃𝑖𝑗𝑛 = 0, ∀ 𝑖

RECURRENT: A state is recurrent if the probability of being revisited from other states is 1. This can happen if
and only if the state is not TRANSIENT.

DOI: 10.35629/4767-08061423 www.ijmsi.org 15 | Page

 Application of Markov Analysis to Consumers Preference

PERIODIC: A state is periodic with period t ≥ 1 if a return is possible only in t, 2t, 3t, ... steps. This means
that P (n) = 0 whenever n is not divisible by t.

Market Equilibrium: The Equilibrium situation is characterized by the equality, 𝑃 = 𝑃. 𝐴. Obviously we

have that;[4]
𝑃1+ 𝑃2 + ⋯ + 𝑃𝑛 = 1.

The state of the market equilibrium, is defined as ”a situation” in which the market shares of the
organization involved in the competition process in the market becomes STEADY-STATE without these shares
unchanged in later periods [4].

VI. METHOD OF DATA COLLECTION

The purpose of this study is to examine the brand loyalty of customers of transport services
using the Markov chains method. Around the main campus of EBSU, there are 3 transport services namely,
Bus, Keke(tricycle) and Bike services. A consumer was defined as a Student who patronize these transport
Services as a means of coming to the campus from their Hostels Frequently on daily basis. For this study,
data of Brand loyalty has been collected from the students of Industrial Mathematics and Applied
Statistics that frequented the Permanent site campus, of Ebonyi State University during the 2018/2019
Academic session. Participants were to fill a questionaire survey anonymously for which 54 consumers
responded and all surveys were fully completed. The survey contained Multiple-Choice questions; Participants
were asked to choose from the 3 transport services, the one they used on that day, select from the same 3 transport
Services, their most preferred and lastly, they were asked to select from (Price, Speed/Time, Comfort and
Availability) the Reason for their choice in the previous question.
Questionnaire survey was used for this study because they are very important in data collection, as they have
major impact on data quality, reliability and validity.

Markov Chain Model has been used to analyze the dynamics (changes) in consumer‟s preference,
predict the Change in preference of a population after a time period (𝑛) and to forecast the state of
equilibrium of the market (Steady-State) after a long-run. As the number of period increases, further changes in
the state Probability become Smaller. Hence, POM-QM for windows software has been used to confirm how
these proportions would Change after each day till they reach Steady-State.
In order to meet our objectives using Markov Chain Method, some Procedures were followed;

Firstly, identify the possible outcomes (𝑖 = 1,2, … , 𝑚) of each of the sequence of events (Transport
Services) in this case 3.Secondly, calculate 𝑃𝑖𝑗 that is the probability of being in State „𝑗‟ in the future given the
current state „i‟ , the outcome 𝑆1 , 𝑆2 , … 𝑆𝑚 are called transition probabilities. When 𝑃𝑖𝑗 = 0, this means that no
transition occurred from state 𝑖 to state 𝑗 and when 𝑃𝑖𝑗 = 1, this means that a system is in state 𝑖 can move only to
State 𝑗 at the next transition. Thirdly, predicting future states involves knowing the system‟s probability of
changing from one state to another. These probabilities were collected and placed in a transition matrix.

Since there are 3 States (transport Services) selected around permanent site campus, the state space is in
the form; S = {Bus, Keke, Bike}. As a result of the survey conducted on the consumers, the data collected about
the relationship between their existing preference and next transport service preferences are transformed into a
markov matrix.

𝐼𝑁𝐼𝑇𝐼𝐴𝐿 𝐵𝑈𝑆 𝐾𝐸𝐾𝐸 𝐵𝐼𝐾𝐸

𝐵𝑈𝑆 𝑋11 𝑎11 𝑎12 𝑎13
𝐾𝐸𝐾𝐸 𝑋21 𝑎21 𝑎22 𝑎23
𝐵𝐼𝐾𝐸 𝑋31 𝑎31 𝑎32 𝑎33

The Markov Matrix of the work

Where;
Column “INITIAL” represents the Present/initial situation (number of customers) of each state (transport
services), and to it‟s right is our transition matrix containing the probability of consumers changing from one
state to another in the future, given the initial state.

To get the first transition, the matrix above was split into matrix A and matrix B
DOI: 10.35629/4767-08061423 www.ijmsi.org 16 | Page
 Application of Markov Analysis to Consumers Preference

where;

𝐼𝑁𝐼𝑇𝐼𝐴𝐿
𝑋11
𝐴= 𝑋21
𝑋31

𝐵𝑈𝑆 𝐾𝐸𝐾𝐸 𝐵𝐼𝐾𝐸

𝑎 𝑎12 𝑎13
𝐵 = 11 .
𝑎21 𝑎22 𝑎23
𝑎31 𝑎32 𝑎33

Then, the transpose of Matrix A was used to multiply matrix B;

𝑎11 𝑎12 𝑎13

𝐴𝑇 𝐵 = 𝑋11 𝑋21 𝑋31 𝑎21 𝑎22 𝑎23
𝑎31 𝑎32 𝑎33

Then the new initial will be;

𝐵𝑈𝑆 𝐾𝐸𝐾𝐸 𝐵𝐼𝐾𝐸

(𝑎11 𝑋11 + 𝑎21 𝑋21 + 𝑎31 𝑋31 ) (𝑎21 𝑋11 + 𝑎22 𝑋21 + 𝑎32 𝑋31 ) (𝑎31 𝑋11 + 𝑎23 𝑋21 + 𝑎33 𝑋31 )

This is the new situation (number of customers patronizing each transport service) of each State (future state) at
t+1.
A new transition matrix will be gotten, by multiplying B by itself, to get Transition Matrix B(2) which satisfies
Chapman-Kolmogorov Equation, as supported by Aypar
& Tuncay(2001).
ij

To reach Equilibrium (Steady State); There was a continual use of each new ”initial” at (t+1) Where t = (1, 2, 3,
...), to multiply each new transition matrix, until we repeatedly get new initials of stationary values and new
transition matrix of the same probability, at which point equilibrium was achieved.

VII. ANALYSIS AND RESULTS

From the survey conducted, out of 54 participants; 28 recently patronized Bus Service, 7 Patronized EBSU
Keke Service and 19 patronized Bike service which amounts to 51.85%, 12.96%, and 35.19% of the population
respectively for each transport service.

Among the 28 consumers who patronized Bus Service; 16 (57.14%) were Satisfied with Bus Services, 8
(28.57%) Preferred Keke Services, and 4 (14.29%) Preferred Bike Services.

Among the 7 consumers who patronized Keke Services; 4 (57.14%) Was Satisfied, only 1 (14.29%) Preferred Bus
Services, 2 (28.57%) Preferred Bike Services.

Among the 19 consumers who patronized Bike Services; 14 (73.68%) were satisfied, 3 (15.79%) Preferred Bus
services, 2 (10.53%) Preferred keke services.

Below is the Representation of the above data in a Transition Matrix form; Events - Choosing a Preferred
transport service.

𝐼𝑁𝐼𝑇𝐼𝐴𝐿 𝐵𝑈𝑆 𝐾𝐸𝐾𝐸 𝐵𝐼𝐾𝐸

𝐵𝑈𝑆 0.5185 0.5714 02857 0.1429 ,
𝐾𝐸𝐾𝐸 0.1296 0.1429 0.5714 0.2857
𝐵𝐼𝐾𝐸 0.3519 0.1579 0.1053 0.7368

DOI: 10.35629/4767-08061423 www.ijmsi.org 17 | Page

 Application of Markov Analysis to Consumers Preference

VIII. MARKET EQUIBRIUM (STEADY STATE)

As earlier stated, solving manually for the steady-state probabilities will involve many iterations. Hence the
use of POM-QM for windows software will be useful; Where states 1, 2, 3 are Bus, Keke and Bike
respectively and Initial contains the probabilities of being in each state at the beginning.

Table 1: Transition Matrix of the work IN POM-QM Format

Table 2: Transition 1 and 2

Ta ble 3: Transition 3 and 4

DOI: 10.35629/4767-08061423 www.ijmsi.org 18 | Page

 Application of Markov Analysis to Consumers Preference

Table 4: Transition 5 and 6

Table 5: Transition 7 and 8

Table 6: Transition 9 and 10

DOI: 10.35629/4767-08061423 www.ijmsi.org 19 | Page

 Application of Markov Analysis to Consumers Preference

Table 7: Transition 11 and 12

Table 8: Transition 13 and 14

Table 9: Transition 15

DOI: 10.35629/4767-08061423 www.ijmsi.org 20 | Page

 Application of Markov Analysis to Consumers Preference

Table10: Final Result Showing Steady-State Probability

IX. DISCUSSIONS
Table 10; The final result, shows the Probabilities at which the market will be in equilibrium
(Steady-state probabilities) which are 0.2619, 0.2857 and 0.4524 for Bus, Keke and Bike respectively. 0.2619 is
the probability of consumer preference changing in favor of Bus, 0.2857 is the probability of consumer
preference changing in favor of keke, while 0.4524 is the probability of consumer preference changing in favor
of Bike.
All our transition matrices are right transition matrices. At Steady-state

0.2619 0.2857 0.4524

The market is said to be in Equilibrium, this state was achieved after 15-iterations (15-days). At the state of
equilibrium, the markov chain becomes absorbing, that is 𝑃𝑖𝑗 = 1 (it will continue to return to itself after any
further iteration (iteration 16, 17, ..., t). This shows that the preference of the consumers will remain constant
and unchanging after Transition 15. Hence, from the first row of the 15th transition matrix

𝐵𝑏 15 𝐾𝑏 15 𝐼𝑏15 = 𝐵𝑏15+1 𝐵𝑏 15+1 𝐵𝑏 15+1

Where,

𝐵𝑏 15 ; The Probability of consumers of EBSU Bus Services remaining Satisfied after 15 iteration (15-days).
𝐾𝑏 15 ; The probability of Bus Service consumers changing their preference to EBSU Keke Services after 15
iterations (15 days).
𝐼𝑏15 ; The probability of Bus Service consumers changing their preference to Bike Services after 15 iterations
(15 days).
Since iteration 15 indicates 15 days, let days = t;

(𝑡) (𝑡) (𝑡)

𝐵𝑏 𝐾𝑏 𝐼𝑏 = 𝐵𝑏 𝑡+1 𝐵𝑏 𝑡+1 𝐵𝑏 𝑡+1

Where; t = (1, 2, 3, ...)

This implies that at equilibrium, the probability of consumer preference for each transport service at time (t),
will remain unchanged in a future time (t+1). The number of customers that will prefer each of the transport
service at Equilibrium (steady state) will be;
(Fraction of consumer preference changing in favour of each transport service)×(Total no. of Consumers)
BUS; 0.2619×54 = 14.1426 ∼ = 14 Consumers,
KEKE; 0.2857 × 54 = 15.66 ∼ = 16 Consumers.
BIKE; 0.4524 × 54 = 24.4296 ∼= 24 Consumers.

In Clearer terms, at equilibrium (long-run);

14(26.19%) of consumers will patronize Bus Services.

16(28.57%) of consumers will patronize Keke Services.
24(45.24%) of consumers will patronize Bike Services.

DOI: 10.35629/4767-08061423 www.ijmsi.org 21 | Page

 Application of Markov Analysis to Consumers Preference

To understand what brought about this change in consumer preference, that has made EBSU Bus Services
fall from being the most patronized to become the least patronized (dynamics of consumer Preference), The
following data were extracted from our initial survey in tabular form.

TRANSPORT SERVICE NO. OF CONSUMERS P S C A P∩A S∩C S∩A

BUS 16 11 1 2 1 1 0 0
KEKE 8 0 3 4 0 0 1 0
BIKE 4 0 3 1 0 0 0 0

Table 11 Initial Consumer Preference of EBSU Bus Service with Respect To some decision variable

TRANSPORT SERVICE NO. OF CONSUMERS P S C A P∩A S∩C S∩A

BUS 1 0 0 1 0 0 0 0
KEKE 4 0 1 1 0 0 1 0
BIKE 2 0 1 1 0 0 0 0
Table12: Initial Consumer Preference of EBSU Keke Service with Respect To some Decision variables

TRANSPORT SERVICE NO. OF CONSUMERS P S C A P∩A S∩C S∩A

BUS 3 1 0 2 0 0 0 0
KEKE 2 0 0 2 0 0 0 0
BIKE 14 0 4 4 3 0 2 1
Table 13: Initial Consumer Preference of Bike Service With Respect To some Decision Variables

TRANSPORT SERVICE NO. OF CONSUMERS P S C A P∩A S∩C S∩A

BUS 28 12 1 5 1 1 0 0
KEKE 7 0 5 7 0 0 0 2
BIKE 19 0 8 6 3 0 1 2
Table 1 4: Summation of Table 11, 12 And 13

The tables contain the following decision variables; Price (P), speed(S), comfort(C), availability(A), price
and availability(P ∩ A), speed and comfort (S ∩ C), speed and availability (S ∩ A) which are defined
below;

PRICE (P): This refers to how affordable and cheap a transport services is.

SPEED(S): This refers to how fast and less time-consuming a transport service is.

COMFORT(C): This refer to space(how less congested a transport service is).

AVAILABILITY (A): This refers to reliability (how easily can a transport service be found around, when
needed urgently).

PRICE AND AVAILABILITY (P ∩ A): This refers to how cheap and reliable a transport service is.

SPEED AND COMFORTABILITY(S ∩ C): These refer to how less time consuming and less congested a
transport service is.

SPEED AND AVAILABILITY(S ∩ A): This refers to how Fast and Reliable a transport service is.

These Variables are what prompted the decision of preference in consumers. Each table shows the
preference of consumers and how many were prompted by each decision variable. As supported by Noor-
ui-Ain Nawaz and Ahmad usman (2016). From table 4 , which sums table 1, 2, and 3. We see that;
22.22%(12) of consumers Preference was influenced by price and 25.93%(14) of consumers preferences was
influenced by speed. 33.33%(18) of consumers preference was influenced by comfort.7.41%(4) of

DOI: 10.35629/4767-08061423 www.ijmsi.org 22 | Page

 Application of Markov Analysis to Consumers Preference

consumers preference was by influenced availability.

1.85%(1) of consumers preference was influenced by both speed and availability. From the above, it is
obvious that consumer preference was mostly influenced by comfort (33.33%), followed by speed, then price
(22.22%). This explains why EBSU Bus Services came out as least patronized at steady-state (Market Equilibrium),
it‟s only advantage was it‟s cheap price, it was not providing enough comfort for its consumers (the number of
consumers per seat (5), is congested and uncomfortable), and it does not have speed (as they waste a lot of time on
the road for example, sorting out balance for consumers in monetary terms). If they can adjust their defaults, they
stand to gain at least 64.81% of consumers (33.33% for speed, 7.41% for speed and comfort, and 1.85% for speed and
availability).
Bike Services at equilibrium rose from second most patronized, to most patronized mostly because they
are fast(less time consuming) and Available. If they maintain this and work on their price, they would remain at the
top of consumer‟s preference list. Keke Services at equilibrium rose from being least patronized to second most
patronized, as it is the most comfortable and second fastest after Bike Services in terms of Speed. There is room for
improvement as they can work on their availability (numbers on the road).

X. CONCLSION
So far we have conducted a Survey of consumer‟s preference on transport services around Ebonyi State
University main campus, the data retrieved was transformed into a Markov matrix and the change in preference of
consumers over a long period of time was observed. After a number of successive transitions on the Markov
matrix, it was possible to reach a balanced condition (Market Equilibrium). According to the mentioned balanced
condition, it was observed that the preferences of consumers intensified on Bike Services and Keke Services, while
that for Bus Services was small. It was also observed that in forming the tendency of changing preference among
consumers, the decision Variables play an important role. In respect to the decision Variables, the ideal transport
Service, will be the one that has the following qualities; comfort, speed, availability, and cheap price, we cannot
generalize for the whole campus, but it is obvious that, to gain the preference of the whole campus, the
transport services have to gain the preference of each department, hence our research paper is very useful on a
micro-scale.

SELECTED REFERENCES
[1]. Amare, B., Estar, A., & Tsehay, S. (April 2018). Application of markov chain analysis model for predicting monthly market
share of restaurants. International Journal Of Recent Engineering Research and Development(IJRERD), 02 , 48 - 55. (ISSN
2455-8761)
[2]. Awogbemi, C. A.,Oloda, F. S., & Caleb, K. (2012 Journal of Economics and Sustainable Development (ISSN 2222-
1700(paper), ISSN 2222-2855(online))
[3]. Hasan, T. (2015). Application of markovian models and transition probabilities matrix to analyze the workforce movement
in jordanian productivity companies statistics (Vol. 4). (ISSN-2250-1991)
[4]. Michael, G. V. (2015). Applications of finite markov chain models to management. Graduate technological Educational
Institute of Western Greece, School of Technological Application .
[5]. Aparna, B., & Sarat, C. K. (2016).IOSR Journal Of Economics and Finance (IOSR-JEF),7, 6-9 (e- ISSN 2321-5933, p-
ISSN:2321-5925)
[6]. Godwin, I. U., Hart,O.A., & Prince,T. E. (2013). Markovian application of: brand switching behaviour A survey of
toothpaste. European Journal Of Business and Management, 05. (ISSN 2222-1905(paper), ISSN 2222-2839(online))
[7]. Hamdy, A. T. (2007). Operations research: An introduction (Eight ed.). Pearson Prentice Hall, Pearson Education Inc.

T. E. Efor, et. al. "Application of Markov Analysis to Consumers Preference." International

Journal of Mathematics and Statistics Invention (IJMSI), vol. 08(06), 2020, pp. 14-23.

DOI: 10.35629/4767-08061423 www.ijmsi.org 23 | Page