0% found this document useful (0 votes)
412 views7 pages

Markov Analysis

Markov analysis is a technique that uses probabilities to forecast future states or occurrences based on current known probabilities. It involves identifying all possible states of a system, determining the probability of a system being in each state, and developing a transition probability matrix to model the likelihood of moving between states. This information can then be used to predict the probabilities of future states. An example is provided of using Markov analysis to predict the future market shares of three grocery stores based on their current shares and historical transition probabilities between stores.

Uploaded by

Sudhir Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
412 views7 pages

Markov Analysis

Markov analysis is a technique that uses probabilities to forecast future states or occurrences based on current known probabilities. It involves identifying all possible states of a system, determining the probability of a system being in each state, and developing a transition probability matrix to model the likelihood of moving between states. This information can then be used to predict the probabilities of future states. An example is provided of using Markov analysis to predict the future market shares of three grocery stores based on their current shares and historical transition probabilities between stores.

Uploaded by

Sudhir Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

Markov Analysis

- It is a technique which forecasts probabilities of future


occurrences by analyzing presently known probabilities
- Technique has numerous applications in business
- Application areas are market share analysis, bad debt
prediction, prediction of machine breakdown
- Markov analysis assumes that system starts in an initial
state/condition
- Two competing manufacturers have 40% and 60% market
share respectively as initial state
- In next two months market share will change to 45% and
55% respectively
- Change in market share can be predicted, if system’s
probability of changing from one state to another is known
- These probabilities ,may be collected or estimated and
presented in a matrix
- This matrix is known as matrix of transition probabilities
- Matrix of transition probabilities presents the likelihood that
the system will change from one period to next
- States indicate all possible conditions of a process/system
- A machine can be any one of the two states at a particular
point of time, i.e., working or breakdown
- A process/system may have many states
- In a small town there are three grocery stores
- A resident can be a customer of any one of the three shops
at a particular point of time
- It presents, there are three states corresponding to three
grocery shops
- In Markov Analysis, it is assumed that the states are both
collectively exhaustive and mutually exclusive
- Collectively exhaustive means all possible states of a
system/process can be identified
- Mutually exclusive means that a system can be in only one
state at any point of time.
- It presents that a resident can be a customer of any one of
the three grocery stores at any point of time
- In case of a problem, first possible states are identified
- Probability that the system is in a particular state is
determined
- Vector of state probabilities is formed
- π (i) = vector of state probabilities for period ‘i’
= (π 1, π 2, π 3,……, π n)
- n = number of states
- π 1, π 2, π 3,……, π n = probability of being in state 1, state 2,
….., state n
- If a machine is working in time period 1, the probability of
working state is 1 and probability of breakdown state is 0
- Vector of state probabilities in period 1 for the machine can
be presented as
- π (1) = (1,0)
- π 1 = 1 = probability of being machine is in first state, i.e.,
working
- π 2 = 0 = probability of being machine is in second state, i.e.,
breakdown
Grocery Store Example
- Three grocery store in a small town
- A resident can be a customer of any one grocery store at a
particular point of time
- Population of the town is 1,00,000
- 40,000 people shop at American Food Store, which will be
termed as State 1
- 30,000 people shop from Food Mart, which will be termed as
State 2
- Remaining 30,000 people are shopping from Atlas Foods,
which will be called as State 3
- Probability that a person will be shopping at one of the three
grocery store are as follows:
- State 1 – American Foods Store – (40000/100000)=0.4
- State 2 – Food Mart – (30000/100000)=0.3
- State 3 – Atlas Foods – (30000/100000)=0.3
- Vector of state probabilities during time period 1 for the
grocery shop example is as follows:
- π (1) = (1,0) = (0.4,0.3, 0.3)
- π (1) = Vector of state probabilities during time period 1 for
the grocery shop
- π 1=0.4=probability that a resident will shop at State 1
- π 2=0.3=probability that a resident will shop at State 2
- π 3=0.3=probability that a resident will shop at State 3
- These probabilities also presents the market share of three
grocery store
- After identifying states and state probabilities, matrix of
transitional probabilities is determined
- Matrix of transitional probabilities along with vector of state
probabilities will be used for predicting future events

Matrix of Transition Probabilities


- Matrix of Transition probabilities allow us to move from
current state to future state
- This is a matrix of conditional probabilities of being in a
future state from a current state
- Let, Pij = conditional probability of being in state ‘j’ in future
from state ‘i’ in current period
- P12 = Probability of being in State 2 in future period, given
that the event is in State 1 in preceding period
- P = Matrix of Transition Probabilities

P11 P12 P13…… P1n


P= P21 P22 P23….. P2n
………………………………….
Pm1 Pm2 Pm3 Pmn

- Pij values are determined empirically


- For example, if it has been observed over time that 10% of
the people currently shopping at Store 1 will be shopping at
Store 2 in next period
- This presents P12 = 0.1
- Matrix of Transition probabilities for three grocery shops are
prepared using historical data

0.8 0.1 0.1


P= 0.1 0.7 0.2
0.2 0.2 0.6

- P11 = 0.8 = probability of being in State 1 after being in State


1 in preceding period
- P12 = 0.1 = probability of being in State 2 after being in State
1 in preceding period
- P13 = 0.1 = probability of being in State 3 after being in State
1 in preceding period
- ………………………………………………………………
- P33 = 0.6 = probability of being in State 3 after being in State
3 in preceding period
Predicting Future Market Share
- State Probabilities for future period may be computed, on
the basis of state probabilities of preceding period and
matrix of transition probabilities
- If current period is 1, state probability for period 2 can be
computed as
- π (2) = π (1) P
- Similarly, state probability for period (n+1) is computed on
the basis of state probabilities of period (n)
- π (n+1) = π (n) P
- Market share for grocery shops for period 2 can be
forecasted as follows:
- π (2) = π (1) P

0.8 0.1 0.1


= (0.4, 0.3, 0.3) 0.1 0.7 0.2
0.2 0.2 0.6

= [(0.4x0.8) + (0.3x0.1) + (0.3x0.2), (0.4x0.1) +


(0.3x0.7) + (0.3x0.2), (0.4x0.1) + (0.3x0.2) + (0.3x0.6)]
= (0.41, 0.31, 0.28)
- It can be observed that American Foods and Food Mart have
improved their market share during period 2
- Market share for Atlas Foods has decreased during period 2
- If this trend continues, does Atlas Foods will lose its total
market share?
- In real life situation if we compute the state probabilities for
number of periods, we will find that the system will come to
equilibrium condition
Example of Machine Operation
- A company has recorded the operation of its milling machine
for several years
- On the basis of past data, vector of state probabilities and
matrix for transition probabilities have been computed
- π (1) = (1,0)

0.8 0.2
P=
0.1 0.9
- P11 = 0.8 = probability that machine will function correctly this
month given that it was functioning correctly last month
- P12 = 0.2 = probability that machine will not function correctly
this month given that it was functioning correctly last month
- P21 = 0.1 = probability that machine will function correctly this
month given that it was not functioning correctly last month
- P22 = 0.9 = probability that machine will not function correctly
this month given that it was not functioning correctly last
month
- π (2) = π (1) P

0.8 0.2
= (1,0)
0.1 0.9
= (0.8, 0.2)
- π (3) = π (2) P

0.8 0.2
= (0.8, 0.2)
0.1 0.9
= (0.66, 0.34)
State Probabilities for the Machine
Period State 1 State 2
1 1.0 0.0
2 0.8 0.2
3 0.66 0.34
4 0.562 0.438
5 0.4934 0.5066
6 0.44538 0.55462
7 0.411766 0.588234
8 0.388236 0.611763
9 0.371765 0.628234
10 0.360235 0.639754
11 0.352165 0.647834
12 0.346515 0.653484
13 0.342560 0.657439
14 0.339792 0.660207
15 0.337854 0.662145
- It can be seen that with time the failure rate of the machine
will come to a equilibrium condition

You might also like