EMDP : Logistics & Supply Chain Management (LSCM-3)

Quantitative Methods Assignment

Prepared by
Anup John Mani
Reg. ID# 87314
PROBABILISTIC DECISION MAKING – is the process that takes into account the “likelihood or chance of something happening” in order to
make decisions which may or may not result in the intended outcome

Example : A certain organization may decide to launch a new product into the market without clear knowledge of whether it will be
successful or not

The key is to understand the data and its distribution. The following 2 concepts are adopted to understand data distribution,
a. Measure of Central Tendency
b. Measure of Dispersion

Measure of Central Tendency – indicates the center point of a dataset and there are 3 common measures of central tendency

i. Mean – is the mathematic average of a dataset. It is calculated by adding all values and then dividing by number of observations.
Mean takes all values in the dataset and change in any one value changes mean. Mean cannot locate center of data accurately in
cases when there are outliers

ii. Median – is the middle value of a dataset that splits the data into half. To find median, data is first sorted from smallest to largest
and then the data point which has equal number of values above and below is located. Locating median is different for dataset
with odd and even values. Outliers have minimal impact on median

iii. Mode – is the value that occurs most frequently in a data set. If the data has more than one value occurring frequently then it will
have multiple modes and if no value repeats then it has no mode
Examples
Measure of Dispersion – indicates how far away the data points are from the central point or mean. In other words it tells us how widely the
data is distributed from the center point, also known as variability. There are 3 common measures of dispersion,

i. Range – of a dataset is the difference between its largest and smallest values. Since range in based on 2 extreme values it is affected
by outliers. If any one of the values is high or low it affects the range

ii. Variance – measures how far each data point in a dataset is from its mean. It is calculated by taking the difference between each
point and the mean, square the difference and then divide the sum of the squares by the total number of data points. Formula for
calculating variance is,
2
σ( 𝑥 − 𝜇)2
𝜎 =
𝑁
where:
• σ2 is the variance
• x is the value of an individual data point
• μ is the mean of data points
• N is the total # of data points

Since calculation is in squared units rather than original units of the data ,variance makes interpreting the data more difficult

iii. Standard Deviation – is the difference between each data point and the mean. In other words it is the square root of the variance. A
smaller standard deviation means that the values are grouped closer to together and vice versa
Example :
Tim wants to order pizza for dinner. He is considering 2 restaurants that specialize in pizza. Both restaurants advertise a mean delivery time
of 25 minutes. Tim’s history for the past 10 orders from both restaurants says differently. He needs to decide which one to place order from
by interpreting the delivery data

Conclusion:
According to the real time data both
restaurants have a mean delivery time of 25
minutes.
R1 has a range of 13 minutes and R2 has
14 minutes. Since range only takes into
account 2 extremes it does not help Tim
arrive at a decision.
The variance of 19 for R1 and 23 for R2
gives me an idea that R2 has a greater
variability compared to R1,but it still does
not tell which one can deliver faster

𝜎𝑅12 =
σ(𝑥1 − 𝜇𝑅1)2
= This is solved by the standard deviation
𝑁𝑅1
which tells Tim that the typical delivery time
𝜎𝑅22 =
σ(𝑥2 − 𝜇𝑅2)2
=
for R2 is approx. +/- 5 minutes from the
𝑁𝑅2 mean while for R1 it is approx. +/- 4
minutes.
Thus Tim can conclude that by ordering from
R1, he has a higher chance of getting the
pizza delivered faster than compared to R2
TRIAL & EXPERIMENT – An experiment is a process of verifying something without knowing what its outcome will be and when an
experiment is repeated several times each one of them is called a Trial. The set of all possible outcomes from an experiment is called
Sample Space. Each outcome from an experiment is called an Event , thus an event can be considered as a subset of sample space.

Example : A Quality Control inspector collects a sample of detergent from the packing line for lab testing. The result of his testing will be
that the detergent is either “good” or “defective”. Based on this we can say that,
Experiment is the act of lab testing the sample detergent
Sample Space of the experiment denoted by S will be “GOOD” and “DEFECTIVE”
Event denoted by E is the outcome of the experiment which can be GOOD or DEFECTIVE

PROBABILITY - The main goal behind experiments is to assign probability to certain events. Probability is a measure of the likelihood that
an event will occur or not. It is reported as number between 0 and 1, where 0 indicates impossibility and 1 indicates certainty. It is
calculated as below

Example : There are 4 blue pens, 5 red pens, 1 yellow pen and 2 orange pens in a box. If you are asked to select 1 pen at random. What
is the probability that the selected pen is yellow or orange
Sample Space (S) = 12 (There 4 + 5 + 1 + 2 = 12 pens in total)
Event (E) = Selected pen is yellow or orange
Number of possible outcomes for event E = # of yellow pens + # of black pens = 1 + 2 = 3

P(E) = P(Yellow or Orange) = 3 / 12 = 0.25

Since the probability 0.25 is closer to 0 we can conclude that the chances of getting a yellow or orange pen is low
RANDOM VARIABLE - is the value of a variable generated from a random experiment and hence is dependent on the outcome of an
experiment. Random variables can be of discrete type where the number of outcomes are known (e.g. number of children in a family)
and continuous type where there are infinite values possible (e.g. the speed of a car)

CUMULATIVE PROBABILITY– refer to the probability that the value of a random variable falls within a range, provided each outcome is not
influenced by the other

Example : If a patient conducts a blood test to check for cholesterol from 3 different labs. What is the probability that the blood test will
result in a “SUCCESS” from not more than 2 labs
Sample Space (S) = 8 (Total of 8 possible outcomes)
Cumulative Probability = P(X<=2) = P(X=0) + P(X=1) + P(X=2) = 0.125 + 0.375 + 0.375 = 0.875

Suppose if X represents the number of failures from this experiment. Then X can take the values 0, 1, 2 or 3. Thus we can say that X is
the Random variable in this example since its value depends on the outcome of the experiment

EXPECTED VALUE (EV) – is the predicted outcome of a trial determined by weighing each possible outcome by the probability of each
outcome occurring. In other words it is the weighted average over all possible outcomes. It is calculated by multiplying each possible
value for the outcome by its probability and then adding these products.
If a trial is repeated several times over a long period, the average of the outcomes from each trial will be closer to the expected value

Example : Company X is looking to expand its retail chain and wants to determine which store to purchase space from. Based on the
study conducted Store A has 60% chance of generating $200,000 in annual profit and 40% chance of generating $250,000. Store B has
70% chance of generating $300,000 and 30% chance of generating $150,000. Which store should company X choose
EV (Store A) = (200,000 * 0.6) + (250,000 * 0.4) = $220,000
EV (Store B) = (300,000 * 0.7) + (150,000 * 0.3) = $255,000
Company X should choose Store B since it has a higher EV
BINOMIAL DISTRIBUTION - is a probability distribution where each trial only has 2 possible outcomes and thus the prefix “Bi” in the name.
Binomial distribution is further characterized by 2 other criteria.
1. The number of trials are fixed
2. The probability of success or failure is the same for each trial
It is calculated using Microsoft Excel as below,

BINOM.DIST ( S, N, P(S), cumulative)

where,
• S is number of successes in trials
• N is number of trials
• P(S) is probability of success for each trial
• cumulative is a value of either TRUE or FALSE. With TRUE it returns cumulative result such as <= or >= and with FALSE it gives result for exact value or ”=“

Example : Market research agency ABC is conducting a survey regarding a specific aspect of customer buying habit. 150 people were
asked “Whether they will buy Product X which is of lower quality at a 50% discount in comparison to Product Y of higher quality but no
discount” . The expected answer is “Yes they will buy” or ”No they will not buy”. The likelihood that a customer will buy Product X is 0.4.
What is the probability that minimum 60 customers will buy Product X and exactly 60 customers will buy Product X

Does this example satisfy the criteria for binomial distribution? YES (explained as below)
1. The number of people to survey is fixed at 150
2. The outcome is whether customer will buy Product X (YES or NO)
3. The probability that each customer will buy Product X is 0.4

S = 60, N = 150, P(S) = 0.4, cumulative = TRUE / FALSE

P (<= than 60 customers will buy X) = BINOM.DIST (60, 150, 0.4, TRUE) = 0.535
P(exactly 60 customers will buy X) = BINOM.DIST (60, 150, 0.4, FALSE) = 0.066
Thus we can say that the chances of minimum of 60 customers buying Product X is 0.535(53.5%) and exactly 60 people buying is
0.066(6.6%)
NORMAL DISTRIBUTION - is a type of probability distribution where the data is symmetric and is clustered around the central value or
mean with no bias to the left or right. When plotted into a graph the data in a normal distribution will take the shape of a symmetrical bell
curve.

MEAN and STANDARD DISTRIBUTION are the 2 key parameters that define the shape and
form of a normal distribution. Due to the symmetrical nature of the data with one peak,
the mean or median or mode divide the data in half or in other words mean = median = mode

EMPIRICAL RULE FOR NORMAL DATA DISTRIBUTION – describes the percentage

of data that fall within specific number of standard deviations from the mean
for normally distributed data
 68% of observations are with +/-1 standard deviation (µ - ∂ to µ + ∂)
 95% are within +/- 2 standard deviations (µ - 2∂ to µ + 2∂)
 99.7% are within +/-3 standard deviations (µ - 3∂ to µ + 3∂)

STANDARD NORMAL DISTRIBUTION – or Z-distribution is a special case of normal distribution where the mean is zero and standard
deviation is 1. The Z value represents the number of standard deviation above or below the mean an observation falls. It is particularly
helpful in understanding where an observation falls relative to other observations in the distribution.
It is calculated as below,
𝑋− 𝜇
Z=
∂
where
• Z is the z-value
• X is observation to be standardized
• 𝜇 is the mean
• ∂ is the standard deviation
NORMAL DISTRIBUTION….continued
It is calculated in Microsoft Excel using the below 2 functions,
NORM.DIST (x, 𝜇, ∂, cumulative)
NORM.S.DIST ( Z, cumulative)
where,
• x is the value for which you need distribution
• 𝜇 is the mean of the distribution
• ∂ is standard deviation of the distribution
• Z is the z-value for the distribution

Example : ABC is the health insurance provider for all employees of company XYZ during the year 2018. ABC is looking at the hospital visit
data for each employee in order to decide on the insurance premium for the new year. The total number of hospital visits per employee is
normally distributed with a mean of 11 visits and standard deviation of 4. As per ABC’s agreement with XYZ if the proportion of
employees visiting hospitals more than 10 times is above 70% then they can increase the insurance premium for next year. Based on the
data available does ABC need to increase the premium for the new year ?

Solution 1:
Z = (10 – 11) / 4 = - 0.25
P (Z>= -0.25) = 1 – P (Z<= -0.25) = 1 – NORM.S.DIST (-0.25, TRUE) = 1 – 0.4013 = 0.5987

Solution 2 :
P (X>=10) = 1– P (X<=10) = 1 - NORM.DIST (10, 11, 4, TRUE) = 1 – 0.4013 = 0.5987

As per the calculation only 59.87% of the employees made more than 10 hospital which does not exceed the threshold stated in the
agreement between ABC and XYZ. Hence ABC does not need to increase the premium for the new year
SAMPLING – In most circumstances the amount of data available on certain business cases is so huge that it becomes difficult to study
them on their own. This huge collection of data is also known as population. In order to gain more insight into the population, samples
are taken to analyze and reach conclusions about the population from which the samples were taken.

CENTRAL LIMIT THEOREM – forms the basis for understanding sample and population distribution. It states that if the size of a sample is
large enough (>=30) then its distribution will be normal irrespective of whether its population is normally distributed or not and the mean
of the sample will be equal to the mean of the population.
Standard deviation and z-value for sample distribution is calculated as below

𝜕 𝑥ҧ − 𝜇
𝜕𝑥ҧ = 𝑍𝑥ҧ =
𝑛 𝜕𝑥ҧ
where,
𝜕 is the population standard deviation
𝜕𝑥ҧ is sample standard deviation
n is the number of samples
𝑥ҧ is the value of sample mean in consideration
𝜇 is population mean
𝑍𝑥ҧ is the z-value for sample distribution

CONFIDENCE INTERVAL – is an interval range of the sample distribution for which there is a specified level of confidence that the
population parameter will fall within this range. Unless specified explicitly the confidence level by default is considered to be at 95%.
Example : As an internal QA/QC auditor for Company XYZ, I am auditing one of its manufacturing units located in India. One of criteria to
be checked is whether the target weight for QC release set by the plant is acceptable or not. I chose 60 bags of Product X at random
produced within the past 6 months for my audit. The QC department assumes a mean weight of 1252 Kg and a standard deviation of 5
Kg.

PART1 - Assuming that the QC department assumption is correct what is the probability the mean weight of my sample is below 1250 Kg

𝜇 = 1252 Kg , 𝜕 = 5 Kg , n = 60 , 𝑥ҧ = 1250 Kg

𝜕 5
𝜕𝑥ҧ = = = 0.645497
𝑛 60
ҧ 𝜇
𝑥− 1250 −1252
𝑍𝑥ҧ = = = -3.098
𝜕𝑥ഥ 0.645497
P(𝑥ҧ <= 1250) = P(𝑍𝑥ҧ <= -3.098) = NORM.S.DIST(-3.098, TRUE) = 0.000974

Based on the results the probability that the mean of my sample is less than 1250 Kg is 0.000974 which is almost equal to zero. This
means that the mean weight assumed by QC department is wrong.

PART2 – Setup a 95% confidence interval for the population mean weight
1250− 𝜇
-1.96 ≤ ≤ +1.96
0.645497

1248.73 ≤ 𝜇 ≤ 1251.27

As per the confidence interval the population mean 1252 Kg falls outside range. Hence the target weight set by the plant is not
acceptable
DETERMINISTIC DECISION MAKING – is the process where business problems are converted to mathematical formulations that help
managers make better decisions. It involves 5 different phases

Phase 1: Problem Definition – Clearly define the problem or in other words define the objective. Incorrect definition of the problem can
result in an undesired solution which could make the problem more worse.
Phase 2 : Model Construction – is creating a visual representation of the problem in form of mathematical relationships or a graph or a
chart.
Phase 3 : Model Solution - The constructed model is then solved using various methods(MS Excel Solver) which in turn solves the problem
at hand
Phase 4 : Model Validation – The constructed model and its solution are then validated to ensure its validity and accuracy for use in future
situations
Phase 5 : Implementation – is the final phase where the model is actually used after it was developed or in other words implementation of
the solution that the model was developed to solve.

Model Construction and Model Solution are the 2 phases where various techniques are utilized to find solution to problems. One such
technique is called Linear programming where we try to find solutions to problems that are subject to restrictions. There are 3 steps to be
followed in applying this technique

Step 1: Define the objective

Step 2 : Identify the restrictions or constraints
Step 3 : Define hidden conditions that are not obvious
Example: ABC Industries produces 2 types of intermediates, X and Y. While X costs $8 to produce, it cost $14 to produce Y. 1 TON of X
yields a profit of $2 and 1 TON of Y yields and profit of $3. ABC does not expect to sell more 2800 TONS in total for both X and Y. The
monthly budget for production costs is limited to $26,200 . ABC wants to determine the best possible production plan for X and Y in order
to get the maximum profit.

OBJECTIVE : Conclusion:
Maximize the profit from sale of X and Y As per the production plan generated by solver ABC can get the maximum
SUBJECT TO : profit if they produce 2167 tons of X and 633 tons of Y
Monthly Budget <= $ 26,200
Monthly Sales <= 2,800 TONS The shadow price indicates that for every $1 increase in monthly budget
the profit will increase by $0.17 and for every 1 ton increase in sales
forecast the profit will increase by $0.67. Allowable increase and decrease
columns tells us the range of change in constraints possible while the
current plan remains optimal.

Final value tells the optimal plan for X and Y. Allowable increase and
decrease columns tells us the range of change possible in objective
coefficients (profit for X & Y) while the current plan remains optimal