Business Statistics and Management Science Notes

Statistical Inference: Hypothesis Testing
1.1 Testing of Hypothesis

Hypothesis testing is a statistical method used to make inferences or draw
conclusions about a population based on sample data. It involves
formulating a null hypothesis (H0) and an alternative hypothesis (Ha). The
null hypothesis represents the status quo or the assumption to be tested,
while the alternative hypothesis represents the researcher's claim or the
alternative to the null hypothesis.
The general steps in hypothesis testing are as follows:
1. State the null hypothesis (H0) and the alternative hypothesis (Ha).
2. Select the level of significance (α), which represents the maximum
probability of making a Type I error (rejecting the null hypothesis
when it is true). Commonly used values for α are 0.05 and 0.01.
3. Collect sample data and calculate the appropriate test statistic.
4. Determine the critical region or rejection region based on the chosen
level of significance and the distribution of the test statistic.
5. Compare the test statistic to the critical value(s) and make a decision.
If the test statistic falls within the critical region, reject the null
hypothesis; otherwise, fail to reject the null hypothesis.
1.2 Testing Hypothesis about the Population Mean and Proportion
1.2.1 Hypothesis Testing for Population Mean (Large Samples)
When testing a hypothesis about the population mean with a large sample
size (typically n > 30), we can use the z-test. The test statistic is calculated
as:
z = (x̄ - μ) / (σ / √n)
where x̄ is the sample mean, μ is the population mean, σ is the population
standard deviation, and n is the sample size. The test statistic follows a
standard normal distribution (z-distribution).
Example:
Suppose we want to test whether the average height of adult males is
significantly different from 175 cm. We collect a sample of 100 male heights
and find the sample mean to be 178 cm, with a known population standard
deviation of 5 cm. We set the significance level at α = 0.05.
Null hypothesis (H0): μ = 175
Alternative hypothesis (Ha): μ ≠ 175
Calculations:
z = (178 - 175) / (5 / √100) = 3 / 0.5 = 6
Page 1 of 74
Since the sample size is large, we can use the standard normal distribution.
The critical values for a two-tailed test with α = 0.05 are approximately
±1.96.
Decision:
Since the calculated test statistic (z = 6) falls in the rejection region (z >
1.96 or z < -1.96), we reject the null hypothesis. There is sufficient evidence
to conclude that the average height of adult males is significantly different
from 175 cm.
1.2.2 Hypothesis Testing for Population Proportion (Large Samples)
When testing a hypothesis about a population proportion, such as the
proportion of successes in a binomial distribution, we use the z-test. The
test statistic is calculated as:
z = (p̂ - p) / √(p(1-p) / n)
where p̂ is the sample proportion, p is the hypothesized population
proportion, and n is the sample size. The test statistic follows a standard
normal distribution (z-distribution).
Example:
Suppose we want to test whether the proportion of defective products
produced by a machine is significantly different from 0.10. We inspect a
sample of 500 products and find that 45 of them are defective. We set the
significance level at α = 0.05.
Null hypothesis (H0): p = 0.10
Alternative hypothesis (Ha): p ≠ 0.10
Calculations:
p̂ = 45/500 = 0.09
z = (0.09 - 0.10) / √[(0.10 * (1-0.10)) / 500] = -0.01 / 0.0141 ≈ -0.71
±1.96.
Decision:
Since the calculated test statistic (z ≈ -0.71) does not fall in the rejection
region (|-0.71| < 1.96), we fail to reject the null hypothesis. There is not
enough evidence to conclude that the proportion of defective products is
significantly different from 0.10.
1.3 Testing Hypothesis about Differences between Two Means or
Proportions
1.3.1 Hypothesis Testing for Differences between Two Means (Independent
Samples)
Page 2 of 74
When testing a hypothesis about the difference between two population
means using independent samples, we can use the t-test. The test statistic
is calculated as:
t = (x̄ 1 - x̄ 2) / √[(s1² / n1)- (s2² / n2)]
where x̄ 1 and x̄ 2 are the sample means, s1 and s2 are the sample standard
deviations, n1 and n2 are the sample sizes. The test statistic follows a t-
distribution with degrees of freedom given by (n1 - 1) + (n2 - 1).
Example:
Suppose we want to test whether there is a significant difference in the
mean weights of two different weight loss programs. We randomly assign
30 individuals to program A and 35 individuals to program B. After the
programs, we record their weights and find the following statistics:
Program A: x̄ 1 = 70 kg, s1 = 5 kg, n1 = 30
Program B: x̄ 2 = 68 kg, s2 = 4.5 kg, n2 = 35
We set the significance level at α = 0.05.
Null hypothesis (H0): μ1 - μ2 = 0
Alternative hypothesis (Ha): μ1 - μ2 ≠ 0
Calculations:
t = (70 - 68) / √[(5² / 30) + (4.5² / 35)] ≈ 2.51
The degrees of freedom for this test are (30 - 1) + (35 - 1) = 63. The critical
value for a two-tailed test with α = 0.05 is approximately ±2.00.
Decision:
Since the calculated test statistic (t ≈ 2.51) falls in the rejection region (|2.51|
> 2.00), we reject the null hypothesis. There is sufficient evidence to
conclude that there is a significant difference in the mean weights of the
two weight loss programs.
1.3.2 Hypothesis Testing for Differences between Two Proportions
(Independent Samples)
When testing a hypothesis about the difference between two population
proportions using independent samples, we can use the z-test. The test
statistic is calculated as:
z = (p̂ 1 - p̂ 2) / √[(p̂ 1 * (1 - p̂ 1) / n1) + (p̂ 2 * (1 - p̂ 2) / n2)]
where p̂ 1 and p̂ 2 are the sample proportions, n1 and n2 are the sample
sizes. The test statistic follows a standard normal distribution (z-
distribution).
Example:
Suppose we want to test whether there is a significant difference in the
Page 3 of 74
proportions of customers who prefer two different brands. We survey 200
customers and find the following results:
Brand A: 120 customers prefer Brand A
Brand B: 75 customers prefer Brand B
We set the significance level at α = 0.05.
Null hypothesis (H0): p1 - p2 = 0
Alternative hypothesis (Ha): p1 - p2 ≠ 0
Calculations:
p̂ 1 = 120/200 = 0.6
p̂ 2 = 75/200 = 0.375
z = (0.6 - 0.375) / √[(0.6 * (1 - 0.6) / 200) + (0.375 * (1 - 0.375) / 200)] ≈ 2.33
±1.96.
Decision:
Since the calculated test statistic (z ≈ 2.33) falls in the rejection region (|
2.33| > 1.96), we reject the null hypothesis. There is sufficient evidence to
conclude that there is a significant difference in the proportions of
customers who prefer the two brands.
1.4 Chi-square Test for Goodness of Fit

The chi-square test for goodness of fit is used to determine whether
observed frequencies within a single categorical variable differ significantly
from the expected frequencies. It is commonly used to assess whether the
distribution of a categorical variable matches a theoretical or expected
distribution.
The general steps for performing a chi-square test for goodness of fit are as
follows:
1. State the null hypothesis (H0) and the alternative hypothesis (Ha). The
null hypothesis assumes that the observed frequencies match the
expected frequencies.
2. Choose the significance level (α) to determine the critical value(s) for
the chi-square test statistic.
3. Collect data and calculate the observed frequencies for each category
of the categorical variable.
4. Determine the expected frequencies for each category based on the
expected distribution or theoretical probabilities.
5. Calculate the chi-square test statistic using the formula:
Page 4 of 74
χ² = Σ [(O - E)² / E]
where O represents the observed frequency and E represents the

expected frequency for each category.
6. Determine the degrees of freedom (df) for the test, which is equal to
the number of categories minus 1.
7. Compare the calculated chi-square test statistic to the critical value(s)
from the chi-square distribution table.
8. If the calculated test statistic exceeds the critical value(s), reject the
null hypothesis and conclude that there is a significant difference
between the observed and expected frequencies.
9. If the calculated test statistic does not exceed the critical value(s), fail
to reject the null hypothesis and conclude that there is no significant
difference between the observed and expected frequencies.
Example:
Suppose we want to test whether the distribution of blood types in a
population follows the expected distribution of 40% A, 30% B, 20% O, and
10% AB. We survey 200 individuals and obtain the following results:
Observed Frequencies:
Blood Type:
A: 80
B: 60
O: 40
AB: 20
Null hypothesis (H0): The distribution of blood types follows the expected
distribution.
Alternative hypothesis (Ha): The distribution of blood types does not follow
the expected distribution.
Calculations:
First, we calculate the expected frequencies based on the expected
distribution:
Expected Frequencies:
A: 0.40 * 200 = 80
B: 0.30 * 200 = 60
O: 0.20 * 200 = 40
AB: 0.10 * 200 = 20
Next, we calculate the chi-square test statistic using the formula:
χ² = [(80 - 80)² / 80] + [(60 - 60)² / 60] + [(40 - 40)² / 40] + [(20 - 20)² / 20]
=0+0+0+0=0
Page 5 of 74
Since there are four categories, the degrees of freedom (df) for the test are
(4 - 1) = 3.
Decision:
To make a decision, we compare the calculated chi-square test statistic to
the critical value from the chi-square distribution table for a given
significance level. Let's assume we choose a significance level of α = 0.05
and consult the chi-square table for df = 3.
From the table, the critical value for α = 0.05 and df = 3 is approximately
7.815.
Since the calculated chi-square statistic (0) is less than the critical value
(7.815), we fail to reject the null hypothesis. There is no significant
difference between the observed and expected frequencies of blood types,
and we conclude that the distribution of blood types in the population
follows the expected distribution.
1.4 Chi-square Test for Independence
The chi-square test for independence is used to determine whether there is
a significant association between two categorical variables. It assesses
whether the distribution of one variable is related to the distribution of
another variable.
The general steps for performing a chi-square test for independence are as
follows:
1. State the null hypothesis (H0) and the alternative hypothesis (Ha). The
null hypothesis assumes that the two categorical variables are
independent.
2. Choose the significance level (α) to determine the critical value(s) for
the chi-square test statistic.
3. Collect data and create a contingency table that shows the
frequencies or counts for each combination of categories for the two
variables.
4. Calculate the expected frequencies for each cell of the contingency
table under the assumption of independence.
5. Calculate the chi-square test statistic using the formula:
χ² = Σ [(O - E)² / E]
where O represents the observed frequency and E represents the

expected frequency for each cell.
Page 6 of 74
6. Determine the degrees of freedom (df) for the test, which is equal to
(r - 1) * (c - 1), where r is the numberof rows (categories) and c is the
number of columns (categories) in the contingency table.
7. Compare the calculated chi-square test statistic to the critical value(s)
from the chi-square distribution table.
8. If the calculated test statistic exceeds the critical value(s), reject the
null hypothesis and conclude that there is a significant association
between the two variables.
9. If the calculated test statistic does not exceed the critical value(s), fail
to reject the null hypothesis and conclude that there is no significant
association between the two variables.
Example:
Suppose we want to test if there is an association between gender (male or
female) and smoking status (smoker, non-smoker) in a sample of 500
individuals. The data collected is summarized in the contingency table
below:
Copy
Smoker Non-smoker
Male 80 120
Female 60 240
Null hypothesis (H0): Gender and smoking status are independent.
Alternative hypothesis (Ha): Gender and smoking status are not
independent.
Calculations:
First, we need to calculate the expected frequencies for each cell under the
assumption of independence. To do this, we calculate the row and column
totals and use them to determine the expected frequencies:
Copy
Smoker Non-smoker Total
Male 80 120 200
Female 60 240 300
Total 140 360 500
Expected Frequencies:
apache
Copy
Smoker Non-smoker
Male (200 * 140) / 500 = 56 (200 * 360) / 500 = 144
Female (300 * 140) / 500 = 84 (300 * 360) / 500 = 216
Next, we calculate the chi-square test statistic:
Copy
χ² = [(80 - 56)² / 56] + [(120 - 144)² / 144] + [(60 - 84)² / 84] + [(240 -
216)² / 216] ≈ 6.29
Page 7 of 74
Since there are two rows and two columns, the degrees of freedom (df) for
the test are (2 - 1) * (2 - 1) = 1.
Decision:
To make a decision, we compare the calculated chi-square test statistic to
the critical value from the chi-square distribution table for a given
significance level. Let's assume we choose a significance level of α = 0.05
and consult the chi-square table for df = 1.
From the table, the critical value for α = 0.05 and df = 1 is approximately
3.841.
Since the calculated chi-square statistic (6.29) is greater than the critical
value (3.841), we reject the null hypothesis. There is a significant association
between gender and smoking status, and we conclude that the two
variables are not independent.
--------------------------------------------------------------------------------------------------------------
Correlation and Regression Analysis

2.1 Simple Regression Analysis
Simple regression analysis is a statistical technique used to determine the
relationship between two variables: a dependent variable (Y) and an
independent variable (X). It helps us understand how changes in the
independent variable affect the dependent variable.
The equation for simple linear regression is:
Y = β0 + β1X + ε
Where:
 Y represents the dependent variable.
 X represents the independent variable.
 β0 represents the intercept (the value of Y when X is equal to zero).
 β1 represents the slope (the change in Y for a one-unit change in X).
 ε represents the error term (the variability in Y that is not explained
by the regression model).
To perform simple regression analysis, we typically follow these steps:
1. State the research question and hypotheses:
 Research question: Is there a significant relationship between X
and Y?
 Null hypothesis (H0): β1 = 0 (There is no relationship between X
and Y).
 Alternative hypothesis (Ha): β1 ≠ 0 (There is a relationship
between X and Y).
Page 8 of 74
2. Collect data: Gather data for both the dependent variable (Y) and
independent variable (X) from a sample or population.
3. Visualize the data: Plot a scatter plot to observe the relationship
between X and Y. This helps identify any patterns or trends.
4. Calculate the regression equation: Use statistical software (such as
SPSS or Excel) to estimate the regression equation coefficients (β0
and β1) that best fit the data.
5. Assess the model fit:
 Coefficient of determination (R²): This measure indicates the
proportion of variability in the dependent variable (Y) that is
explained by the independent variable (X). R² ranges from 0 to
1, with higher values indicating a better fit.
 Adjusted R²: Similar to R², but it penalizes the inclusion of
irrelevant predictors in multiple regression models.
6. Evaluate the significance of the regression equation:
 Conduct a hypothesis test to determine whether the slope
coefficient (β1) is significantly different from zero.
 Calculate the standard error of the estimate to assess the
precision of the regression line.
7. Interpret the results:
 Examine the coefficients: β0 (intercept) and β1 (slope) provide
insights into the relationship between X and Y.
 Interpret the statistical significance: If the slope coefficient is
significantly different from zero, we can conclude that there is a
significant relationship between X and Y.
Example:
Suppose we want to investigate the relationship between the number of
hours studied (X) and the exam score (Y) for a group of students. We
collected data from 20 students and obtained the following results:
json
Copy
Number of Hours Studied (X): [5, 7, 3, 2, 6, 8, 4, 10, 9, 7, 1, 3, 6, 9, 8, 2, 5,
4, 10, 7]
Exam Score (Y): [65, 80, 55, 50, 70, 90, 60, 95, 85, 75, 45, 55, 70,
85, 80, 50, 65, 60, 90, 75]
Step 1: State the research question and hypotheses:
 Research question: Is there a significant relationship between the
number of hours studied and the exam score?
 Null hypothesis (H0): β1 = 0 (There is no relationship between the
number of hours studied and the exam score).
Page 9 of 74
 Alternative hypothesis (Ha): β1 ≠ 0 (There is a relationship between
the number of hours studied and the exam score).
Step 2: Collect data (already done).
Step 3: Visualize the data:
 Plot a scatter plot of the number of hours studied (X) versus the exam
score (Y).
Step 4: Calculate the regression equation:
 Use statistical software (e.g., SPSS) to estimate the regression
equation coefficients (β0 and β1).
Assuming the regression equation is:
Y = β0 + β1X + ε
The estimated regression equation might be:
Y = 35.7 + 6.9X
Step 5: Assess the model fit:
 Calculate the coefficient of determination (R²) to determine the
proportion of the variability in the exam score (Y) explained by the
number of hours studied (X).
Assuming R² = 0.75, it means that 75% of the variability in the exam score is
explained by the number of hours studied.
Step 6: Evaluate the significance of the regression equation:
 Perform a hypothesis test to determine whether the slope coefficient
(β1) is significantly different from zero.
Assuming the p-value for β1is less than 0.05, we reject the null hypothesis
and conclude that there is a significant relationship between the number of
hours studied and the exam score.
Step 7: Interpret the results:
 The intercept (β0 = 35.7) suggests that if a student doesn't study at
all (X = 0), their expected exam score would be 35.7.
 The slope (β1 = 6.9) indicates that for every additional hour studied,
the expected exam score increases by 6.9 points.
In conclusion, based on the analysis of the data, there is a significant
positive relationship between the number of hours studied and the exam
score. Students who study more hours tend to achieve higher exam scores.
2.2 Multiple Regression Analysis

Multiple regression analysis is a statistical technique used to understand
the relationship between a dependent variable (Y) and multiple
independent variables (X₁, X₂, X₃, ..., Xₚ). It allows us to examine how each
Page 10 of 74
independent variable contributes to the variation in the dependent variable
while controlling for the effects of other variables.
The equation for multiple linear regression is:
Y = β₀ + β₁X₁ + β₂X₂ + β₃X₃ + ... + βₚXₚ + ε
Where:
 Y represents the dependent variable.
 X₁, X₂, X₃, ..., Xₚ represent the independent variables.
 β₀ represents the intercept.
 β₁, β₂, β₃, ..., βₚ represent the coefficients associated with each
independent variable.
 ε represents the error term.
To perform multiple regression analysis, we typically follow these steps:
1. State the research question and hypotheses:
 Research question: What is the relationship between the
independent variables (X₁, X₂, X₃, ..., Xₚ) and the dependent
variable (Y)?
 Null hypothesis (H0): β₁ = β₂ = β₃ = ... = β ₚ = 0 (None of the
independent variables have a significant relationship with the
dependent variable).
 Alternative hypothesis (Ha): At least one of the independent
variables (X₁, X₂, X₃, ..., Xₚ) has a significant relationship with the
dependent variable.
2. Collect data: Gather data for the dependent variable (Y) and multiple
independent variables (X₁, X₂, X₃, ..., Xₚ) from a sample or population.
3. Visualize the data: Use scatter plots, correlation matrices, or other
graphical methods to observe the relationships between the
independent variables and the dependent variable. This helps identify
any patterns or correlations.
4. Calculate the regression equation: Use statistical software (such as
SPSS) to estimate the regression equation coefficients (β₀, β₁, β₂, β₃, ...,
βₚ) that best fit the data.
5. Assess the model fit:
 Coefficient of determination (R²): This measure indicates the
proportion of variability in the dependent variable (Y) that is
explained by the independent variables (X₁, X₂, X₃, ..., X ₚ). R²
ranges from 0 to 1, with higher values indicating a better fit.
Page 11 of 74
 Adjusted R²: Similar to R², but it takes into account the number
of predictors and penalizes the inclusion of irrelevant
predictors.
6. Evaluate the significance of the regression equation:
 Conduct hypothesis tests to determine whether the coefficients
(β₁, β₂, β₃, ..., βₚ) are significantly different from zero.
 Calculate the standard errors of the estimates to assess the
precision of the coefficients.
7. Interpret the results:
 Examine the coefficients (β₁, β₂, β₃, ..., β ₚ) to understand the
direction and strength of the relationships between the
independent variables and the dependent variable.
 Interpret the statistical significance of the coefficients: If a
coefficient is significantly different from zero, we can conclude
that the corresponding independent variable has a significant
relationship with the dependent variable.
Multiple regression analysis enables us to predict the value of the
dependent variable based on the values of the independent variables, as
well as understand the relative importance of each independent variable in
explaining the variation in the dependent variable.
Example:
Suppose we want to understand the factors influencing the sales revenue
(Y) of a company. We collected data from 50 different stores and
considered the following independent variables: advertising expenditure
(X₁), store size (X₂), and location (X₃). We obtained the following results:
json
Copy
Advertising Expenditure (X₁): [1000, 1500, 800, 1200, 900, 1100, 1400, 1300,
1000, 950, 1250, 1050, 1150, 1350, 850, 950, 1100, 1300, 1500, 900, 1050, 950,
1200, 1100, 1300, 1000, 900, 950, 1200, 1000, 1100, 1400, 1300, 1000, 950, 1250,
1050, 1150, 1350, 850, 950, 1100, 1300, 1500, 900, 1050, 950, 1200, 1100, 1300]
Store Size (X₂): [500, 600, 480, 550, 520, 530, 590, 570, 510, Sure!
Let's continue the example:
Store Size (X₂): [500, 600, 480, 550, 520, 530, 590, 570, 510, 540, 560, 550,
570, 520, 530, 590, 570, 510, 540, 560, 550, 570, 520, 530, 590, 570, 510,
540, 560, 550, 570, 520, 530, 590, 570, 510, 540, 560, 550, 570, 520, 530,
590, 570, 510, 540, 560, 550, 570]
Location (X₃): [1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2,
3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3, 1, 2, 3]
Sales Revenue (Y): [25000, 28000, 23000, 26000, 24000, 27000, 29000,
Page 12 of 74
28000, 25000, 24000, 27000, 26000, 28000, 29000, 23000, 24000, 27000,
29000, 28000, 25000, 24000, 23000, 26000, 27000, 29000, 25000, 24000,
23000, 26000, 25000, 27000, 29000, 28000, 25000, 24000, 27000, 26000,
28000, 29000, 23000, 24000, 27000, 29000, 28000, 25000, 24000, 23000,
26000, 27000, 29000]
pgsql
Copy
In this example, we have 50 observations from 50 different stores. The independent
variables are advertising expenditure (X₁), store size (X₂), and location (X₃),
while the dependent variable is the sales revenue (Y).
To perform multiple regression analysis using SPSS, follow these steps:
1. Launch SPSS and open a new data file.

2. Enter the data into SPSS by creating separate columns for each variable (X₁,
X₂, X₃, Y) and inputting the corresponding values.
3. Click on "Analyse" in the top menu, then select "Regression" and choose
"Linear."
4. Move the dependent variable (Y) into the "Dependent" box and the independent
variables (X₁, X₂, X₃) into the "Independent(s)" box.
5. Optional: Click on the "Statistics" button to select additional statistics you
want to compute, such as R², adjusted R², and standardized coefficients.
6. Click "OK" to run the multiple regression analysis.
SPSS will generate output that includes the regression coefficients, standard
errors, t-values, p-values, and additional statistics.
Interpreting the results:
1. Coefficients: The coefficient values represent the estimated effect of each

independent variable on the dependent variable while holding other variables
constant. For example:
- β₀ (intercept): The estimated sales revenue when all independent variables

are zero.
- β₁ (advertising expenditure): The estimated change in sales revenue for a
one-unit increase in advertising expenditure, holding other variables constant.
- β₂ (store size): The estimated change in sales revenue for a one-unit
increase in store size, holding other variables constant.
- β₃ (location): The estimated change in sales revenue for a one-unit increase
in location, holding other variables constant.
2. Standard Errors: These values provide an indication of the precision of the

coefficient estimates. Smaller standard errors indicate greater precision.
3. t-values and p-values: These values are used to test the null hypothesis that
the corresponding coefficient is equal to zero. If the p-value is below a
specified significance level (e.g., 0.05), we can reject the null hypothesis and
conclude that the independent variable has a significant relationship with the
dependent variable.
4. R² and Adjusted R²: These measures indicate the proportion of variability in

the dependent
Page 13 of 74
3.1 Methodology of Operations Research / Management Science
Operations Research (OR), also known as Management Science, is a
multidisciplinary approach that uses mathematical and analytical methods
to solve complex problems and make better decisions in organizations. The
methodology of Operations Research involves the following steps:
1. Problem Formulation: Clearly define the problem and identify the
decision variables, constraints, and objectives. This step involves
understanding the problem context, gathering relevant data, and
specifying the desired outcomes.
2. Model Development: Create a mathematical or computational model
that represents the problem situation. The model should capture the
relationships between decision variables, constraints, and objectives.
Various modeling techniques, such as optimization, simulation,
queuing theory, and network analysis, can be used.
3. Data Collection and Analysis: Collect and analyze the relevant data
required for the model. This may involve gathering historical data,
conducting surveys, or using existing databases. Data analysis
techniques, such as statistical analysis, regression, and forecasting,
are applied to gain insights and validate the model.
4. Model Solution: Use appropriate solution methods to solve the
formulated model. Optimization techniques, such as linear
programming, integer programming, dynamic programming, and
heuristic algorithms, are commonly employed. Simulation models can
be executed to generate results in a stochastic environment.
5. Model Validation: Validate the model's accuracy and reliability by
comparing its results with real-world observations or expert
judgment. Sensitivity analysis and robustness testing are performed
to assess the model's performance under different scenarios and
parameter variations.
6. Decision Making: Interpret the results of the model and provide
recommendations to decision-makers. These recommendations
should consider the trade-offs between conflicting objectives, the
impact of constraints, and the practicality of implementing the
suggested solutions.
7. Implementation and Monitoring: Implement the recommended
solutions and monitor their performance over time. Evaluate the
Page 14 of 74
effectiveness of the implemented solutions and make adjustments if
necessary.
The methodology of Operations Research combines mathematical
modeling, optimization techniques, statistical analysis, and computer
simulation to provide quantitative insights and support decision-making
processes in various fields, including manufacturing, logistics, finance,
healthcare, and transportation.
3.2 Successful Application of Management Science / Operations
Research
Management Science or Operations Research has been successfully applied
in numerous areas to improve decision-making and optimize processes.
Some notable successful applications include:
1. Supply Chain Management: Operations Research techniques are used
to optimize inventory management, distribution network design,
transportation routing, and demand forecasting. These applications
help minimize costs, reduce lead times, and improve customer service
levels.
2. Production Planning and Scheduling: Operations Research models aid
in optimizing production schedules, resource allocation, and capacity
planning. By considering factors such as machine availability, labor
constraints, and production costs, organizations can achieve efficient
and balanced production processes.
3. Project Management: Operations Research techniques help in project
scheduling, resource allocation, and project cost estimation. Critical
Path Method (CPM) and Program Evaluation and Review Technique
(PERT) are commonly used to manage complex projects and ensure
timely completion.
4. Financial Planning and Portfolio Optimization: Operations Research
models assist in portfolio selection, asset allocation, and risk
management. By considering factors such as expected returns, risk
tolerance, and correlation among assets, optimal investment
strategies can be determined.
5. Healthcare Management: Operations Research techniques are applied
to optimize hospital resource allocation, patient scheduling,
emergency room management, and healthcare facility location. These
applications aim to improve patient outcomes, reduce waiting times,
and optimize resource utilization.
Page 15 of 74
6. Transportation and Logistics: Operations Research models help
optimize route planning, vehicle routing, facility location, and freight
consolidation. These applications minimize transportation costs,
improve delivery efficiency, and reduce carbon emissions.
7. Energy Management: Operations Research techniques aid in
optimizing energy generation, distribution, and consumption. They
help identify optimal energy mix, plan renewable energy integration,
and optimize energy pricing and demand response programs.
These are just a few examples of the wide-ranging applications of
Operations Research and Management Science. The field continues to
evolve and find new ways to improve efficiency, reduce costs, and enhance
decision-making in various industries.
3.3 What is Linear Programming?

Linear Programming (LP) is a mathematical optimization technique used to
find the best possible solution to a problem with linear constraints and a
linear objective function. In LP, the objective is to maximize or minimize a
linear mathematical expression (the objective function) subject to a set of
linear constraints.
The general form of a linear programming problem can be expressed as
follows:
Maximize (or Minimize):
Z = c₁x₁ + c₂x₂ + ... + cₙxₙ
Subject to:
a₁₁x₁ + a₁₂x₂ + ... + a₁ₙxₙ ≤ b₁
a₂₁x₁ + a₂₂x₂ + ... + a₂ₙxₙ ≤ b₂
...
aₘ₁x₁ + aₘ₂x₂ + ... + aₘₙxₙ ≤ bₘ
x₁, x₂, ..., xₙ ≥ 0
Where:
 Z represents the objective function, which is either maximized or
minimized.
 c₁, c₂, ..., cₙ are the coefficients of the decision variables x₁, x₂, ..., x ₙ in
the objective function.
 a₁₁, a₁₂, ..., aₘₙ are the coefficients of the decision variables in the
constraints.
 b₁, b₂, ..., bₘ are the right-hand side values of the constraints.
 m represents the number of constraints.
Page 16 of 74
 n represents the number of decision variables.
The decision variables x₁, x₂, ..., xₙ represent the quantities to be
determined, while the constraints limit the values that these variables can
take. The goal is to find values for the decision variables that satisfy all the
constraints and optimize the objective function.
Linear programming can be solved using various algorithms, such as the
Simplex Method, Interior Point Methods, and the Dual Simplex Method.
These algorithms iteratively explore the feasible region defined by the
constraints to find the optimal solution.
3.4 The Graphical Solution of a 2-Variable LP
The graphical solution method is a visual approach to solving a 2-variable
linear programming problem. It involves graphing the constraints and
objective function on a coordinate plane to determine the feasible region
and identify the optimal solution.
To solve a 2-variable LP graphically, follow these steps:
Step 1: Formulate the LP problem.
Z = c₁x₁ + c₂x₂
Subject to:
a₁₁x₁ + a₁₂x₂ ≤ b₁
a₂₁x₁ + a₂₂x₂ ≤ b₂
x₁, x₂ ≥ 0
Step 2: Graph the constraints.
For each constraint, rewrite it in the form of an equation (ignoring the
inequality sign) and plot the corresponding line on the coordinate plane. To
do this, solve the equation for one variable and choose values for the other
variable to plot the line. Repeat this process for each constraint.
Step 3: Identify the feasible region.
The feasible region is the area where all constraints are satisfied
simultaneously. It is the intersection of the shaded regions formed by the
constraint lines.
Step 4: Determine the optimal solution.
a) Identify the feasible region. It may be unbounded, bounded, or empty.
b) Evaluate the objective function at the corner points (vertices) of the
feasible region. These corner points represent the extreme solutions.
c) If the objective is to maximize, select the corner point with the highest
objective function value as the optimal solution. If the objective is to
minimize, select the corner point with the lowest objective function value.
Page 17 of 74
Step 5: Interpret the solution.
The optimal solution represents the values of the decision variables that
maximize or minimize the objective function while satisfying all the
constraints. The coordinates of the optimal solution point (x₁*, x₂*)
correspond to the values of x₁ and x₂ that yield the optimal solution.
Example:
Maximize Z = 3x₁ + 4x₂
Subject to:
2x₁ + x₂ ≤ 8
x₁ + 2x₂ ≤ 6
x₁, x₂ ≥ 0
Graphical Solution:
Step 1: Formulate the LP problem and constraints.
Subject to:
2x₁ + x₂ ≤ 8
x₁ + 2x₂ ≤ 6
Step 2: Graph the constraints.
To graph the first constraint, rewrite it as an equation:
2x₁ + x₂ = 8
Choose values for x₁ and solve for x₂:
When x₁ = 0, 1x₂ = 8 → x₂ = 8
When x₂ = 0, 2x₁ = 8 → x₁ = 4
Plot the points (0, 8) and (4, 0) and connect them toform the line
representing the first constraint.
To graph the second constraint, rewrite it as an equation:
x₁ + 2x₂ = 6
Choose values for x₁ and solve for x₂:
When x₁ = 0, 2x₂ = 6 → x₂ = 3
When x₂ = 0, x₁ = 6
Plot the points (0, 3) and (6, 0) and connect them to form the line
representing the second constraint.
Step 3: Identify the feasible region.
The feasible region is the area where both constraints are satisfied
simultaneously. It is the shaded region formed by the intersection of the
two constraint lines.
Step 4: Determine the optimal solution.
Page 18 of 74
In this example, the feasible region is bounded and has four corner points:
(0, 0), (0, 3), (2, 2), and (4, 0).
Evaluate the objective function at each corner point:
Z₁ = 3(0) + 4(0) = 0
Z₂ = 3(0) + 4(3) = 12
Z₃ = 3(2) + 4(2) = 14
Z₄ = 3(4) + 4(0) = 12
Since the objective is to maximize, the optimal solution is (2, 2) with Z = 14.
The optimal solution (2, 2) represents the values of x₁ and x₂ that maximize
the objective function Z = 3x₁ + 4x₂ while satisfying the constraints. The
maximum value of the objective function is 14.
3.5 The Simultaneous Equation Approach to a 2-Variable LP

The simultaneous equation approach is an algebraic method used to solve
a 2-variable linear programming problem. It involves setting up a system of
equations using the constraints and solving them simultaneously to find the
optimal solution.
To solve a 2-variable LP using the simultaneous equation approach, follow
these steps:
Step 1: Formulate the LP problem.
Z = c₁x₁ + c₂x₂
Subject to:
a₁₁x₁ + a₁₂x₂ ≤ b₁
a₂₁x₁ + a₂₂x₂ ≤ b₂
x₁, x₂ ≥ 0
Step 2: Convert the inequality constraints to equations.
For each inequality constraint, replace the inequality sign with an equal sign
to obtain equations.
Step 3: Set up the system of equations.
Write the equations obtained from the constraints in the form:
a₁₁x₁ + a₁₂x₂ = b₁
a₂₁x₁ + a₂₂x₂ = b₂
Step 4: Solve the system of equations.
Solve the system of equations simultaneously to find the values of x₁ and x₂
that satisfy all the constraints. This can be done using various methods,
such as substitution, elimination, or matrix algebra.
Page 19 of 74
Step 5: Evaluate the objective function.
Evaluate the objective function Z = c₁x₁ + c₂x₂ using the values of x₁ and x₂
obtained in the previous step to determine the optimal solution.
The optimal solution represents the values of the decision variables x₁ and
x₂ that maximize or minimize the objective function while satisfying all the
constraints.
Example:
Subject to:
2x₁ + x₂ ≤ 8
x₁ + 2x₂ ≤ 6
Simultaneous Equation Approach:
Step 1: Formulate the LP problem and constraints.
Subject to:
2x₁ + x₂ ≤ 8
x₁ + 2x₂ ≤ 6
Step 2: Convert the inequality constraints to equations.
2x₁ + x₂ = 8
x₁ + 2x₂ = 6
Step 3: Set up the system of equations.
2x₁ + x₂ = 8
x₁ + 2x₂ = 6
Step 4: Solve the system of equations.
Using the substitution method:
From the second equation, x₁ = 6 - 2x₂
Substitute this value of x₁ into the first equation:
2(6 - 2x₂) + x₂ = 8
12 - 4x₂ + x₂ = 8
-3x₂ = -4
x₂ = 4/3
Substitute this value of x₂ back into the second equation:
x₁ + 2(4/3) = 6
x₁ + 8/3 = 6
x₁ = 6 - 8/3
x₁ = 10/3
Step 5: Evaluate the objective function.
Page 20 of 74
Substitute the values of x₁ and x₂ into the objective function:
Z = 3(10/3) + 4(4/3)
Z = 10 + 16/3
Z = 46/3
The optimal solution is x₁ = 10/3 and x₂ = 4/3, which maximize the
objective function Z = 46/3. These values represent the quantities of x₁ and
x₂ that yield the optimal solution.
3.6 Work Scheduling Problem
The work scheduling problem is a type of optimization problem that
involves assigning tasks to workers or machines while considering various
constraints and objectives. The goal is to find the most efficient schedule
that minimizes the total time or cost required to complete the tasks.
In a work scheduling problem, the following elements are typically
considered:
1. Tasks: A set of tasks that need to be completed. Each task has specific
requirements, such as duration, dependencies on other tasks, and
resource constraints.
2. Workers or Machines: The available resources responsible for
performing the tasks. Each worker or machine may have different
capabilities, availability, and capacity constraints.
3. Constraints: Various constraints that must be satisfied, such as task
dependencies, resource availability, and precedence relationships.
4. Objective Function: The objective of the scheduling problem, which
can be minimizing the total time required to complete all tasks,
minimizing the cost, or maximizing the utilization of resources.
Solving a work scheduling problem typically involves formulating it as an
optimization model, which can be linear programming, integer
programming, or other optimization techniques. The objective function and
constraints are defined based on the specific requirements and objectives
of the problem.
Example:
Consider awork scheduling problem with the following tasks and
constraints:
Tasks:
1. Task A: Duration = 5 hours, requires Worker 1.
2. Task B: Duration = 3 hours, requires Worker 2.
3. Task C: Duration = 4 hours, requires either Worker 1 or Worker 2.
Page 21 of 74
4. Task D: Duration = 6 hours, requires Worker 1.
Constraints:
1. Worker 1 can work a maximum of 8 hours per day.
2. Worker 2 can work a maximum of 6 hours per day.
3. Task C can only be assigned to one worker.
Objective:
Minimize the total time required to complete all tasks.
To solve this work scheduling problem, we can formulate it as a linear
programming problem.
Decision Variables:
Let x₁ be the number of hours Worker 1 works.
Let x₂ be the number of hours Worker 2 works.
Objective Function:
Minimize Z = 5x₁ + 3x₂
Constraints:
1. x₁ + x₂ ≥ 5 (Task A requires 5 hours)
2. x₁ + x₂ ≥ 3 (Task B requires 3 hours)
3. x₁ + x₂ ≥ 4 (Task C requires 4 hours)
4. x₁ + x₂ ≥ 6 (Task D requires 6 hours)
5. x₁ ≤ 8 (Worker 1 maximum work hours)
6. x₂ ≤ 6 (Worker 2 maximum work hours)
7. x₁, x₂ ≥ 0
Solving this linear programming problem will provide the optimal solution,
which represents the number of hours each worker should work to
minimize the total time required to complete all tasks, while satisfying the
constraints.
3.7 Blending Problems

Blending problems involve finding the optimal combination of ingredients
or components to achieve a desired outcome while considering various
constraints. These problems are commonly encountered in industries such
as manufacturing, agriculture, and food processing, where different
ingredients or components need to be mixed to meet specific
requirements.
In a blending problem, the following elements are typically considered:
1. Ingredients or Components: A set of ingredients or components with
different properties or characteristics. Each ingredient may have a
cost, availability, and specific attributes.
Page 22 of 74
2. Constraints: Various constraints that must be satisfied, such as
minimum or maximum quantities of ingredients, nutritional or
chemical composition requirements, and availability of ingredients.
3. Objective Function: The objective of the blending problem, which can
be minimizing the cost, maximizing the quality, or achieving specific
target values for certain attributes.
To solve a blending problem, follow these steps:
Step 1: Define the decision variables.
Identify the quantities of each ingredient or component to be used in the
blend. Assign a decision variable to each ingredient, representing the
quantity of that ingredient in the blend.
Step 2: Formulate the objective function.
Define the objective function based on the desired outcome. This could be
minimizing the cost, maximizing the quality, or achieving specific target
values for certain attributes. The objective function is typically a linear
combination of the decision variables and may involve ingredient costs,
quality measures, or other relevant factors.
Step 3: Specify the constraints.
Identify the constraints that need to be satisfied. These constraints could
include minimum or maximum quantities of ingredients, nutritional or
chemical composition requirements, availability constraints, or other
specific limitations. Each constraint should be formulated as an equation or
inequality involving the decision variables.
Step 4: Solve the optimization problem.
Solve the formulated optimization problem using techniques such as linear
programming, integer programming, or nonlinear programming,
depending on the complexity of the problem and the nature of the
constraints. The solution will provide the optimal values for the decision
variables that satisfy the constraints and optimize the objective function.
Step 5: Analyze the solution and interpret the results.
Examine the solution obtained and interpret the values of the decision
variables. These values represent the optimal combination of ingredients or
components that achieve the desired outcome while satisfying the
constraints. Analyze the results to gain insights into the composition of the
blend and its characteristics.
Example:
Consider a blending problem for producing a nutritional supplement with
the following ingredients and constraints:
Page 23 of 74
Ingredients:
1. Ingredient A: Cost = $5/kg, contains 25% protein, and requires a
minimum quantity of 100 kg.
2. Ingredient B: Cost = $8/kg, contains 20% protein, and requires a
3. Ingredient C: Cost = $10/kg, contains 15% protein, and requires a
Constraints:
1. The total quantity of the supplement should be at least 200 kg.
2. The protein content in the supplement should be at least 18%.
3. The cost of the supplement should be minimized.
To solve this blending problem, we can formulate it as a linear
Decision Variables:
Let x₁ be the quantity (in kg) of Ingredient A in the supplement.
Let x₂ be the quantity (in kg) of Ingredient B in the supplement.
Let x₃ be the quantity (in kg) of Ingredient C in the supplement.
Objective Function:
Minimize Z = 5x₁ + 8x₂ + 10x₃
Constraints:
1. x₁ + x₂ + x₃ ≥ 200 (Total quantity constraint)
2. (0.25x₁ + 0.20x₂ + 0.15x₃) / (x₁ + x₂ + x₃) ≥ 0.18 (Protein content
constraint)
3. x₁ ≥ 100 (Minimum quantity constraint for Ingredient A)
4. x₂ ≥ 50 (Minimum quantity constraint for Ingredient B)
5. x₃ ≥ 75 (Minimum quantity constraint for Ingredient C)
6. x₁, x₂, x₃ ≥ 0
Solving this linear programming problem will provide the optimal solution,
which represents the quantities of each ingredient to be used in the
nutritional supplement to minimize the cost while satisfying the constraints.
3.8 Capital Budgeting Problems
Capital budgeting problems involve making investment decisions regarding
long-term projects or assets. These problems require analyzing the cash
flows associated with different investment alternatives and assessing their
financial viability.
In capital budgeting, the following elements are typically considered:
1. Investment Options: A set of investment alternatives or projects with
different cash flows, costs, and expected returns. Each investment
Page 24 of 74
option represents a potential long-term commitment of financial
resources.
2. Cash Flows: The cash inflows and outflows associated with each
investment option over its lifetime. Cash flows include initial
investments, expected revenues or savings, operating costs, taxes,
salvage values, and any other relevant financial flows.
3. Time Value of Money: TheTime value of money: The concept that
states that a dollar received in the future is worth less than a dollar
received today due to the opportunity cost of not having that money
available for investment or consumption. Time value of money is
considered by discounting future cash flows to their present value
using an appropriate discount rate.
4. Evaluation Criteria: Criteria used to assess the financial viability and
profitability of investment options. Common evaluation criteria
include net present value (NPV), internal rate of return (IRR), payback
period, and profitability index.
To solve a capital budgeting problem, follow these steps:
Step 1: Define the investment options.
Identify and describe the investment alternatives or projects under
consideration. Each investment option should be clearly defined in terms of
its cash flows, costs, and expected returns.
Step 2: Estimate cash flows.
Estimate the cash inflows and outflows associated with each investment
option over its lifetime. This includes initial investments, expected revenues
or savings, operating costs, taxes, salvage values, and any other relevant
financial flows. Cash flows should be estimated for each period and
adjusted for the time value of money.
Step 3: Determine the appropriate discount rate.
Select an appropriate discount rate that reflects the risk and opportunity
cost associated with the investment. The discount rate is used to calculate
the present value of future cash flows, taking into account the time value of
money.
Step 4: Apply evaluation criteria.
Apply the chosen evaluation criteria, such as net present value (NPV),
internal rate of return (IRR), payback period, or profitability index, to assess
the financial viability and profitability of each investment option. Calculate
the values of these criteria based on the estimated cash flows and the
chosen discount rate.
Page 25 of 74
Step 5: Make the investment decision.
Compare and analyze the results obtained from the evaluation criteria for
each investment option. Consider other factors such as strategic alignment,
risk analysis, and qualitative aspects. Make the investment decision based
on the findings and select the option that best aligns with the
organization's goals and financial objectives.
Step 6: Monitor and review.
After making the investment, monitor the performance of the chosen
option and regularly review its financial outcomes against the initial
projections. Adjust the investment strategy if necessary based on the actual
results and changing market conditions.
Example:
Suppose a company is considering two investment options, Option A and
Option B, with the following estimated cash flows:
Option A:
Initial Investment: $100,000
Cash Flows: Year 1: $30,000, Year 2: $50,000, Year 3: $40,000, Year 4:
$20,000, Year 5: $10,000
Option B:
Initial Investment: $150,000
Cash Flows: Year 1: $40,000, Year 2: $40,000, Year 3: $40,000, Year 4:
$40,000, Year 5: $40,000
The company uses a discount rate of 10% to account for the time value of
money.
To evaluate these investment options using the net present value (NPV)
criterion, we calculate the present value of each cash flow using the
discount rate and sum them up.
Option A NPV:
NPV = -$100,000 + $30,000/(1+0.10) + $50,000/(1+0.10)^2 +
$40,000/(1+0.10)^3 + $20,000/(1+0.10)^4 + $10,000/(1+0.10)^5
Option B NPV:
NPV = -$150,000 + $40,000/(1+0.10) + $40,000/(1+0.10)^2 +
$40,000/(1+0.10)^3 + $40,000/(1+0.10)^4 + $40,000/(1+0.10)^5
By calculating the NPV for each option, we can compare the values and
determine which investment option is more financially viable. A positive
NPV indicates that the investment generates more cash inflows than
outflows and is considered favorable.
Page 26 of 74
In addition to NPV, other evaluation criteria such as internal rate of return
(IRR), payback period, and profitability index can also be applied to assess
the investment options and make a well-informed decision.
-------------------------------------------------------------------------------------------------------
4.1 Formulating Transportation Problems

Transportation problems involve finding the optimal allocation of goods or
resources from a set of sources to a set of destinations, considering
constraints such as supply, demand, and transportation costs. These
problems commonly arise in logistics and supply chain management, where
efficient allocation and distribution of goods are crucial.
In a transportation problem, the following elements are typically
considered:
1. Sources: A set of locations or suppliers where goods are available for
shipment. Each source has a limited supply quantity.
2. Destinations: A set of locations or customers where goods need to be
delivered. Each destination has a specific demand quantity.
3. Transportation Costs: The costs associated with shipping goods from
sources to destinations. Transportation costs can vary depending on
the distance, mode of transport, and other factors.
4. Supply and Demand Constraints: Constraints that limit the supply
quantity from sources and the demand quantity at destinations.
These constraints must be satisfied in the optimal allocation.
To solve a transportation problem, follow these steps:
Step 1: Define decision variables.
Identify the decision variables that represent the quantity of goods to be
transported from each source to each destination. Assign a variable to each
combination of source and destination, representing the quantity to be
shipped.
Define the objective function based on the desired outcome. This could be
minimizing the total transportation cost, maximizing the total profit, or
minimizing the total distance traveled. The objective function is typically a
linear combination of the decision variables and the transportation costs.
Identify the constraints that need to be satisfied. These constraints include
supply constraints (total supply from each source) and demand constraints
(total demand at each destination). Additionally, there may be constraints
Page 27 of 74
on the quantities shipped from each source and received at each
destination. Each constraint should be formulated as an equation or
Solve the formulated optimization problem using techniques such as the
transportation simplex method, the northwest corner method, or the
Vogel's approximation method. These methods iteratively improve the
allocation until the optimal solution is reached. The solution will provide the
optimal allocation of goods from sources to destinations that minimizes the
transportation cost or achieves the desired objective.
variables. These values represent the optimal quantity of goods to be
transported from each source to each destination. Analyze the results to
gain insights into the allocation pattern, identify bottlenecks or
inefficiencies, and make informed decisions for improving the
transportation system.
Example:
Consider a transportation problem where there are three sources (S1, S2,
S3) and four destinations (D1, D2, D3, D4). The supply and demand
quantities, along with the transportation costs, are given as follows:
Supply:
S1: 100 units
S2: 150 units
S3: 200 units
Demand:
D1: 80 units
D2: 120 units
D3: 100 units
D4: 150 units
Transportation costs (in dollars per unit):
apache
Copy
D1 D2 D3 D4
S1 | 4 6 9 11
S2 | 5 7 8 10
S3 | 6 8 7 9
To solve this transportation problem, we can formulate it as a linear
Page 28 of 74
Decision Variables:
Let xᵢⱼ be the quantity of goods to be transported from source Sᵢ to
destination Dⱼ.
Objective Function:
Minimize Z = 4x₁₁ + 6x₁₂ + 9x₁₃ + 11x₁₄ + 5x₂₁ + 7x₂₂ + 8x₂₃ + 10x₂₄ + 6x₃₁
+ 8x₃₂ + 7x₃₃ + 9x₃₄
Constraints:
1. x₁₁ + x₁₂ + x₁₃ + x₁₄ ≤ 100 (Supply constraint for S1)
2. x₂₁ + x₂₂ + x₂₃ + x₂₄ ≤ 150 (Supply constraint for S2)
3. x₃₁ + x₃₂ + x₃₃ + x₃₄ ≤ 200 (Supply constraint for S3)
4. x₁₁ + x₂₁ + x₃₁ ≥ 80 (Demand constraint for D1)
5. x₁₂ + x₂₂ + x₃₂ ≥ 120 (Demand constraint for D2)
6. x₁₃ + x₂₃ + x₃₃ ≥ 100 (Demand constraint for D3)
7. x₁₄ + x₂₄ + x₃₄ ≥ 150 (Demand constraint for D4)
8. All decision variables xᵢⱼ ≥ 0 (Non-negativity constraint)
Solving this linear programming problem will provide the optimal allocation
of goods from sources to destinations, minimizing the total transportation
cost.
4.2 Assignment Problems

Assignment problems involve the allocation of a set of resources or tasks to
a set of individuals or machines, considering certain criteria such as cost,
efficiency, or preference. These problems commonly arise in various fields,
including operations research, personnel management, and project
scheduling.
In an assignment problem, the key elements considered are:
1. Resources/Tasks: A set of resources or tasks that need to be assigned.
Each resource or task has specific characteristics, requirements, or
values.
2. Individuals/Machines: A set of individuals or machines available for
assignment. Each individual or machine has the capacity, capability,
or preference for performing certain tasks or handling specific
resources.
3. Assignment Costs/Values: The costs or values associated with
assigning a resource/task to an individual/machine. These costs or
values can represent various criteria, such as monetary cost, time
taken, efficiency, or preference.
Page 29 of 74
4. Assignment Constraints: Constraints that limit the assignment
possibilities. For example, a resource/task can only be assigned to
one individual/machine, and an individual/machine can only be
assigned to one resource/task.
To solve an assignment problem, the following steps can be followed:
Identify the decision variables that represent the assignment of
resources/tasks to individuals/machines. Assign a binary variable to each
combination of resource/task and individual/machine, representing
whether the assignment is made or not.
Define the objective function based on the desired outcome. This could
involve minimizing the total assignment cost, maximizing the total
assignment value, or achieving a specific objective based on the problem
context. The objective function is typically a linear combination of the
decision variables and the assignment costs or values.
Identify the constraints that need to be satisfied. These constraints typically
include ensuring that each resource/task is assigned to exactly one
individual/machine and that each individual/machine is assigned to at most
one resource/task. Each constraint should be formulated as an equation or
Hungarian algorithm or the auction algorithm. These methods efficiently
find the optimal assignment by iteratively improving the assignment until
the optimal solution is reached. The solution will provide the optimal
assignment of resources/tasks to individuals/machines that minimizes the
assignment cost or achieves the desired objective.
variables. These values represent the optimal assignment of resources/tasks
to individuals/machines. Analyze the results to gain insights into the
assignment pattern, identify any constraints or limitations, and make
informed decisions based on the assignment solution.
Example:
Consider an assignment problem where there are four tasks (T1, T2, T3, T4)
and four individuals (I1, I2, I3, I4). The assignment costs are given as follows:
Page 30 of 74
Assignment Costs:
apache
Copy
I1 I2 I3 I4
T1 | 4 6 9 11
T2 | 5 7 8 10
T3 | 6 8 7 9
T4 | 3 5 4 6
To solve this assignment problem, we can formulate it as a linear
Decision Variables:
Let xᵢⱼ be a binary variable representing the assignment of task Tᵢ to
individual Iⱼ. xᵢⱼ = 1 if the assignment is made, and xᵢ ⱼ = 0 otherwise.
Objective Function:
Minimize Z = 4x₁₁ + 6x₁₂ + 9x₁₃ + 11x₁₄ + 5x₂₁ + 7x₂₂ + 8x₂₃ + 10x₂₄ + 6x₃₁
+ 8x₃₂ + 7x₃₃ + 9x₃₄ + 3x₄₁ + 5x₄₂ + 4x₄₃ + 6x₄₄
Constraints:
1. x₁₁ + x₁₂ + x₁₃ + x₁₄ = 1 (Task assignment constraint for T1)
2. x₂₁ + x₂₂ + x₂₃ + x₂₄ = 1 (Task assignment constraint for T2)
3. x₃₁ + x₃₂ + x₃₃ + x₃₄ = 1 (Task assignment constraint for T3)
4. x₄₁ + x₄₂ + x₄₃ + x₄₄ = 1 (Task assignment constraint for T4)
5. x₁₁ + x₂₁ + x₃₁ + x₄₁ = 1 (Individual assignment constraint for I1)
6. x₁₂ + x₂₂ + x₃₂ + x₄₂ = 1 (Individual assignment constraint for I2)
7. x₁₃ + x₂₃ + x₃₃ + x₄₃ = 1 (Individual assignment constraint for I3)
8. x₁₄ + x₂₄ + x₃₄ + x₄₄ = 1 (Individual assignment constraint for I4)
All decision variables are binary: xᵢⱼ ∈ {0, 1}
Solving this linear programming problem will give the optimal assignment
of tasks to individuals that minimizes the total assignment cost. The
solution will provide the values of the decision variables (xᵢ ⱼ) indicating the
assignment status.
4.3 Transhipment Problems

Transhipment problems are a type of optimization problem that involve the
movement of goods or resources from multiple sources to multiple
destinations, passing through intermediate locations called transhipment
points. These problems commonly arise in logistics, supply chain
management, and distribution network optimization.
In a transhipment problem, the main elements considered are:
1. Sources: A set of locations or nodes where goods/resources originate.
Page 31 of 74
2. Transhipment Points: Intermediate locations or nodes through which
goods/resources pass during transportation.
3. Destinations: A set of locations or nodes where goods/resources are
delivered.
4. Supply and Demand: The supply capacity or availability of
goods/resources at the sources and the demand or requirement of
goods/resources at the destinations.
5. Transportation Costs: The costs associated with transporting
goods/resources between various nodes, including the costs of
transportation from sources to transhipment points and from
transhipment points to destinations.
6. Flow Variables: The flow variables represent the quantity of
goods/resources transported between nodes. These variables indicate
the amount of flow from sources to transhipment points, from
transhipment points to destinations, or directly from sources to
destinations.
To solve a transhipment problem, the following steps can be followed:
Identify the decision variables that represent the flow of goods/resources
between nodes. Assign a variable to each combination of nodes, indicating
the quantity or flow between them.
Define the objective function based on the desired outcome. This could
involve minimizing the total transportation cost, maximizing the total flow,
or achieving a specific objective based on the problem context. The
objective function is typically a linear combination of the decision variables
and the transportation costs.
Identify the constraints that need to be satisfied. These constraints include
ensuring that the flow of goods/resources is conserved at each node,
satisfying the supply and demand requirements, and enforcing any
additional constraints specific to the problem context. Each constraint
should be formulated as an equation or inequality involving the decision
variables.
transportation simplex method or network flow algorithms like the
minimum cost flow algorithm. These methods efficiently find the optimal
Page 32 of 74
flow of goods/resources between nodes that minimizes the transportation
cost or achieves the desired objective. The solution will provide the optimal
flow values for the decision variables.
variables. These values represent the optimal flow of goods/resources
between nodes. Analyze the results to gain insights into the flow patterns,
identify any bottlenecks or inefficiencies, and make informed decisions
based on the optimal flow solution.
Example:
Consider a transhipment problem with the following network:
Copy
D1 (Demand: 30)
/ | \
Source S1 (Supply: 20) T1 D2 (Demand: 40)
\ | /
S2 (Supply: 30)
The transportation costs between the nodes are given as follows:
gherkin
Copy
Transportation Costs:
T1
| 3 4 |
S1 | 5 2 | D1
| 6 1 |
-----------------
S2 | 2 3 | D2
| 4 5 |
To solve this transhipment problem, we can formulate it as a linear
Decision Variables:
Let xᵢⱼ represent the flow of goods/resources from node i to node j.
Objective Function:
Minimize Z = 3x₁₁ + 4x₁₂ + 5x₁₃ + 2x₂₁ + 6x₂₂ + x₂₃ + 6x₃₁ + x₃₂ + 2x₃₃ + 4x₄₁
+ 5x₄₂
Constraints:
Supply Constraints:
1. x₁₁ + x₁₂ + x₁₃ ≤ 20 (Supply constraint for S1)
2. x₂₁ + x₂₂ + x₂₃ + x₃₁ + x₃₂ + x₃₃ + x₄₁ + x₄₂ ≤ 30 (Supply constraint for
S2)
Demand Constraints:
3. x₁₃ + x₂₃ + x₃₃ + x₄₃ ≥ 30 (Demand constraint for D1)
4. x₁₂ + x₂₂ + x₃₂ + x₄₂ + x₄₁ ≥ 40 (Demand constraint for D2)
Page 33 of 74
Flow Conservation Constraints:
5. x₁₁ + x₂₁ = x₁₂ + x₁₃ (Flow conservation at T1)
x₃₁ + x₄₁ = x₃₂ + x₃₃ (Flow conservation at T1)
Non-negativity Constraints:
7. xᵢⱼ ≥ 0 (For all decision variables)
By solving this linear programming problem, we can find the optimal flow
values that minimize the transportation cost.
Once the problem is solved, the optimal flow values can be used to make
decisions regarding the transportation of goods/resources between nodes.
The solution may reveal the most cost-effective routes, identify any
bottlenecks, or provide insights for optimizing the flow in the network.
--------------------------------------------------------------------------------------------------------------
5.1 Decision Criteria

Decision making under uncertainty involves making choices when the
outcomes and probabilities of different events are not known with certainty.
In such situations, decision criteria are used to evaluate alternative courses
of action and select the most appropriate one. Here are some commonly
used decision criteria:
1. Maximax Criterion: This criterion selects the alternative that
maximizes the maximum possible outcome. It assumes an optimistic
decision maker who seeks to maximize potential gains. It does not
consider the probabilities of events.
Scenario:
You are a project manager considering three potential projects: Project A,
Project B, and Project C. Each project has a different potential profit, and
you want to use the Maximax criterion to determine which project to
choose, focusing on maximizing the best possible outcome.
Here are the profit outcomes for each project:
Project A:
 Profit Outcome 1: $50,000
Project B:
Page 34 of 74
Project C:
To apply the Maximax criterion, we select the project that maximizes the
best possible outcome (i.e., the maximum profit).
Maximax analysis:
For Project A, the maximum profit is $80,000 (Outcome 2).
For Project B, the maximum profit is $70,000 (Outcome 1).
For Project C, the maximum profit is $90,000 (Outcome 1).
Based on the Maximax criterion, we choose the project with the highest
potential profit. In this case, Project C has the maximum profit of $90,000,
which is higher than the maximum profits of Projects A and B.
Therefore, according to the Maximax criterion, Project C is the preferred
choice as it offers the highest potential profit. This decision focuses on
maximizing the best possible outcome without considering the
probabilities or the risks associated with each project.
It's important to note that the Maximax criterion does not consider the
likelihood of each outcome occurring or any potential risks involved.
Therefore, this criterion may ignore important factors such as project
feasibility, resource requirements, or the likelihood of achieving the
maximum profit.
2. Maximin Criterion: This criterion selects the alternative that maximizes

the minimum possible outcome. It assumes a pessimistic decision
maker who seeks to minimize potential losses. It does not consider
the probabilities of events.
Scenario:
You are a business owner considering two potential suppliers for a critical
component in your manufacturing process: Supplier A and Supplier B. Each
supplier has different quality levels, and you want to use the Maximin
criterion to determine which supplier to choose, focusing on minimizing the
worst possible outcome.
Here are the quality ratings for each supplier:
Supplier A:
 Quality Rating 1: 8
Page 35 of 74
Supplier B:
To apply the Maximin criterion, we select the supplier that minimizes the
worst possible outcome (i.e., the minimum quality rating).
Maximin analysis:
For Supplier A, the minimum quality rating is 6 (Rating 2).
For Supplier B, the minimum quality rating is 4 (Rating 3).
Based on the Maximin criterion, we choose the supplier with the highest
minimum quality rating. In this case, Supplier A has a minimum quality
rating of 6, which is higher than the minimum quality rating of Supplier B
(4).
Therefore, according to the Maximin criterion, Supplier A is the preferred
choice as it offers a higher minimum quality rating. This decision focuses on
minimizing the worst possible outcome without considering the
probabilities or the potential for higher quality ratings.
It's important to note that the Maximin criterion does not consider the
likelihood of each quality rating occurring or any potential risks involved.
Therefore, this criterion may not capture the full picture of supplier
performance or other important factors such as pricing, reliability, or
customer support.
3. Maximizing Expected Monetary Value (EMV): This criterion calculates

the expected monetary value by multiplying each possible outcome
by its corresponding probability and selecting the alternative with the
highest expected value. It considers both the outcomes and their
probabilities.
Example
You are a project manager considering two potential projects: Project A and
Project B. Each project has a different probability of success and associated
monetary value, and you want to use the Maximizing EMV criterion to
determine which project to choose, focusing on maximizing the expected
monetary value.
Page 36 of 74
Here are the details of the projects:
Project A:
 Probability of Success: 0.7
 Monetary Value if Successful: $100,000
 Monetary Value if Unsuccessful: $0
Project B:
 Monetary Value if Unsuccessful: $20,000
To apply the Maximizing EMV criterion, we calculate the expected monetary

value for each project and select the one with the highest value.
Maximizing EMV analysis:
For Project A:
EMV(A) = Probability(Success) * Monetary Value(Success) +
Probability(Failure) * Monetary Value(Failure)
= 0.7 * $100,000 + 0.3 * $0
= $70,000
For Project B:
EMV(B) = Probability(Success) * Monetary Value(Success) +
= 0.5 * $150,000 + 0.5 * $20,000
= $85,000
Based on the Maximizing EMV criterion, we choose the project with the
highest expected monetary value. In this case, Project B has an EMV of
$85,000, which is higher than the EMV of Project A ($70,000).
Therefore, according to the Maximizing EMV criterion, Project B is the
preferred choice as it offers a higher expected monetary value. This
decision focuses on maximizing the potential monetary outcome by
considering both the probabilities of success and failure.
It's important to note that the Maximizing EMV criterion assumes that the
monetary values and probabilities are accurate and reliable. However, it
may not capture other important factors such as project feasibility, strategic
alignment, or non-monetary considerations.
Page 37 of 74
4. Minimizing Expected Monetary Value (EMV): This criterion is the
opposite of the maximizing EMV. It selects the alternative with the
lowest expected value. It is suitable when the decision maker wants to
minimize potential costs or losses.
Example
Project B. Each project has a different probability of success and associated
monetary value, and you want to use the Minimizing EMV criterion to
determine which project to choose, focusing on minimizing the expected
monetary value.
Here are the details of the projects:
Project A:
 Monetary Value if Unsuccessful: $0
Project B:
 Monetary Value if Unsuccessful: $50,000
To apply the Minimizing EMV criterion, we calculate the expected monetary
value for each project and select the one with the lowest value.
Minimizing EMV analysis:
For Project A:
EMV(A) = Probability(Success) * Monetary Value(Success) +
= 0.8 * $200,000 + 0.2 * $0
= $160,000
For Project B:
EMV(B) = Probability(Success) * Monetary Value(Success) +
= 0.6 * $150,000 + 0.4 * $50,000
= $120,000
Based on the Minimizing EMV criterion, we choose the project with the
lowest expected monetary value. In this case, Project B has an EMV of
$120,000, which is lower than the EMV of Project A ($160,000).
Therefore, according to the Minimizing EMV criterion, Project B is the
preferred choice as it offers a lower expected monetary value. This decision
Page 38 of 74
focuses on minimizing the potential monetary outcome by considering
both the probabilities of success and failure.
5. Minimax Regret Criterion: This criterion focuses on the regret

associated with each alternative. Regret is the difference between the
maximum payoff for a particular state of nature and the actual payoff
received. The minimax regret criterion selects the alternative that
minimizes the maximum regret across all states of nature.
Example
You are a business owner considering three potential locations for opening
a new store: Location A, Location B, and Location C. Each location has a
different potential profit, and you want to use the Minimax Regret Criterion
to determine which location to choose, focusing on minimizing the
maximum regret.
Here are the profit outcomes for each location:
Location A:
Location B:
Location C:
To apply the Minimax Regret Criterion, we calculate the regret values for
each location and select the one with the lowest maximum regret.
Minimax Regret analysis:
To calculate the regret values, we subtract each profit outcome from the
maximum profit for each outcome. Then, we take the maximum regret
value for each location and select the location with the lowest maximum
regret.
For Location A:
Regret Outcome 1 = $90,000 - $50,000 = $40,000
Page 39 of 74
Regret Outcome 2 = $90,000 - $80,000 = $10,000
Regret Outcome 3 = $90,000 - $60,000 = $30,000
Maximum Regret for Location A = $40,000
For Location B:
Regret Outcome 1 = $90,000 - $70,000 = $20,000
Regret Outcome 2 = $90,000 - $55,000 = $35,000
Regret Outcome 3 = $90,000 - $65,000 = $25,000
Maximum Regret for Location B = $35,000
For Location C:
Regret Outcome 1 = $90,000 - $90,000 = $0
Regret Outcome 2 = $90,000 - $75,000 = $15,000
Regret Outcome 3 = $90,000 - $70,000 = $20,000
Maximum Regret for Location C = $20,000
Based on the Minimax Regret Criterion, we choose the location with the
lowest maximum regret. In this case, Location A has a maximum regret of
$40,000, Location B has a maximum regret of $35,000, and Location C has a
maximum regret of $20,000.
Therefore, according to the Minimax Regret Criterion, Location C is the
preferred choice as it offers the lowest maximum regret. This decision
focuses on minimizing the potential regret by considering the difference
between the actual profit outcomes and the maximum profit for each
location.
It's important to note that the Minimax Regret Criterion assumes that the
profit outcomes are accurate and reliable. However, it may not capture
other important factors such as location-specific advantages, market
conditions, or non-monetary considerations.
6. Hurwicz Criterion: This criterion involves a coefficient of optimism (α)

that represents the decision maker's attitude towards risk. It
calculates a weighted average of the maximum outcome and the
minimum outcome, based on the coefficient of optimism. The
alternative with the highest weighted average is selected.
Example
Project B. Each project has a different potential payoff, and you want to use
Page 40 of 74
the Hurwicz Criterion to determine which project to choose, focusing on
balancing optimism and pessimism.
Here are the payoffs for each project:
Project A:
 Payoff if Successful: $100,000
 Payoff if Unsuccessful: $20,000
 Optimism Factor (α): 0.6
Project B:
 Payoff if Successful: $150,000
 Payoff if Unsuccessful: $10,000
 Optimism Factor (α): 0.4
To apply the Hurwicz Criterion, we calculate the weighted average of the
best and worst payoffs for each project, using the optimism factor (α) to
strike a balance between optimism and pessimism.
Hurwicz Criterion analysis:
For Project A:
Hurwicz Score (A) = α * Payoff(Success) + (1 - α) * Payoff(Failure)
= 0.6 * $100,000 + 0.4 * $20,000
= $76,000
For Project B:
Hurwicz Score (B) = α * Payoff(Success) + (1 - α) * Payoff(Failure)
= 0.4 * $150,000 + 0.6 * $10,000
= $64,000
Based on the Hurwicz Criterion, we choose the project with the highest
Hurwicz score. In this case, Project A has a Hurwicz score of $76,000, and
Project B has a Hurwicz score of $64,000.
Therefore, according to the Hurwicz Criterion, Project A is the preferred
choice as it has a higher Hurwicz score. This decision strikes a balance
between optimism and pessimism, giving more weight to the potential
payoff if successful while still considering the potential payoff if
unsuccessful.
It's important to note that the Hurwicz Criterion relies on the choice of the
optimism factor (α), which determines the balance between optimism and
pessimism. Different values of α can lead to different decisions.
Additionally, the Hurwicz Criterion does not consider the probabilities of
success or failure, making it less suitable for decision-making when
probabilities are available.
Page 41 of 74
7. Expected Utility Criterion: This criterion takes into account the
decision maker's preferences or utility function for different
outcomes. It assigns a utility value to each outcome and calculates
the expected utility of each alternative by multiplying the utility of
each outcome by its corresponding probability. The alternative with
the highest expected utility is selected.
Scenario:
You are a decision-maker considering two potential investment options:
Option A and Option B. Each option has different potential returns and
associated probabilities, and you want to use the Expected Utility Criterion
to determine which option to choose, focusing on maximizing expected
utility.
Here are the details of the investment options:
Option A:
 Return if Successful: $100,000
 Return if Unsuccessful: -$10,000
 Probability of Failure: 0.4
Option B:
 Return if Successful: $120,000
 Return if Unsuccessful: -$5,000
 Probability of Failure: 0.6
To apply the Expected Utility Criterion, we calculate the expected utility for
each option and select the one with the highest value.
Expected Utility analysis:
For Option A:
Expected Utility (A) = Probability(Success) * Utility(Return if Successful) +
Probability(Failure) * Utility(Return if Unsuccessful)
Utility(Return if Successful) = Return if Successful
Utility(Return if Unsuccessful) = Return if Unsuccessful
Expected Utility (A) = 0.6 * $100,000 + 0.4 * (-$10,000)
= $60,000 + (-$4,000)
= $56,000
Page 42 of 74
For Option B:
Expected Utility (B) = Probability(Success) * Utility(Return if Successful) +
Probability(Failure) * Utility(Return if Unsuccessful)
Utility(Return if Successful) = Return if Successful
Utility(Return if Unsuccessful) = Return if Unsuccessful
Expected Utility (B) = 0.4 * $120,000 + 0.6 * (-$5,000)
= $48,000 + (-$3,000)
= $45,000
Based on the Expected Utility Criterion, we choose the option with the
highest expected utility. In this case, Option A has an expected utility of
$56,000, and Option B has an expected utility of $45,000.
Therefore, according to the Expected Utility Criterion, Option A is the
preferred choice as it has a higher expected utility. This decision focuses on
maximizing the expected utility, which incorporates both the potential
returns and the decision-maker's utility function.
5.2 Utility Theory
Utility theory is a decision-making framework that incorporates the

preferences of decision makers under conditions of uncertainty. It assumes
that decision makers have subjective utility functions that assign values
representing preferences to different outcomes. Key concepts in utility
theory include:
1. Utility Function: A utility function maps outcomes to utility values,
representing the decision maker's preferences. It quantifies the
desirability or satisfaction associated with different outcomes. Utility
functions are typically increasing, reflecting the principle of
diminishing marginal utility.
2. Expected Utility: The expected utility of an alternative is calculated by
multiplying the utility of each possible outcome by its corresponding
probability and summing these values. It provides a measure of the
average utility that can be expected from choosing a particular
alternative.
3. Risk Aversion and Risk Seeking: Decision makers' attitudes towards
risk are reflected in their utility functions. Risk-averse decision makers
have concave utility functions, indicating diminishing marginal utility
for gains. Risk-seeking decision makers have convex utility functions,
Page 43 of 74
indicating increasing marginal utility for gains. Risk neutrality
corresponds to a linear utility function.
4. Certainty Equivalent: The certainty equivalent is the guaranteed
outcome that a decision maker would consider equally preferable to
the uncertain outcome. It represents the value at which the decision
maker is indifferent between the certain outcome and the uncertain
outcome.
Utility theory provides a framework for decision makers to evaluate and
compare alternatives based on their expected utility. By incorporating
individual preferences, it allows decision makers to make choices that align
with their risk attitudes and priorities.
Scenario:
You are considering two investment options: Option A and Option B. Each
option has a different potential return and associated probability. You want
to use utility theory to determine which option is more desirable based on
your risk preferences.
Here are the details of the investment options:
Option A:
 Potential Return: $10,000
 Probability: 0.6
Option B:
 Potential Return: $15,000
 Probability: 0.3
To apply utility theory, we need to assign utility values to the potential
returns. Let's assume a utility function that reflects risk aversion:
Utility($10,000) = 0.8
Utility($15,000) = 0.6
Now, let's calculate the expected utility for each option:
Expected Utility of Option A:
EU(A) = Utility($10,000) * Probability(A)
= 0.8 * 0.6
= 0.48
Expected Utility of Option B:
EU(B) = Utility($15,000) * Probability(B)
= 0.6 * 0.3
= 0.18
Page 44 of 74
Based on expected utility, the higher the value, the more desirable the
option. In this case, Option A has an expected utility of 0.48, while Option B
has an expected utility of 0.18.
Therefore, according to utility theory, Option A is more desirable than
Option B due to its higher expected utility. This decision reflects your risk
aversion, as Option A has a lower potential return but a higher probability,
resulting in a more favorable expected utility.
The utility theory relies on the assumption that individuals make decisions
based on the expected utility of outcomes. The specific utility function used
and the associated values are subjective and may vary depending on an
individual's risk preferences and personal circumstances.
5.3 Decision Trees

Decision trees are graphical representations of decision problems that
involve sequential decision making and uncertainty. They are composed of
nodes and branches and are particularly useful for analyzing decisions with
multiple stages and probabilistic outcomes. Key components of decision
trees include:
1. Decision Nodes: Decision nodes represent points in the decision-
making process where the decision maker must choose between
different alternatives or actions. These nodes are usually depicted as
squares or rectangles.
2. Chance Nodes: Chance nodes represent uncertain events or states of
nature that occur after a decision is made. They are depicted as
circles or ovals and are connected to branches representing different
possible outcomes.
3. Branches: Branches represent the different alternatives or outcomes
at each decision or chance node. They show the different paths or
sequences of events that can occur.
4. Payoff or Outcome Nodes: Payoff nodes represent the final outcomes
or payoffs associated with specific sequences of decisions and chance
events. They are depicted as triangles and typically include the
associated values or utilities.
Decision trees are constructed by starting with the initial decision node and
sequentially adding chance nodes and their corresponding branches. The
probabilities and values associated with each chance node and payoff node
are assigned based on available information or estimates.
Page 45 of 74
Decision trees allow decision makers to visually analyze and evaluate the
different alternatives and their expected outcomes. By calculating the
expected values or utilities at the payoff nodes, decision makers can
identify the optimal decision path or alternative that maximizes their
expected utility or value.
Example
You are considering whether to go on a weekend trip or stay at home. Your
decision will depend on the weather forecast and whether you have
completed all your pending tasks.
Here is the decision tree:
Start
/ \
Sunny Rainy
/ \
Tasks Done Tasks Not Done
/ \
Go on Trip Stay at Home
In this decision tree:
 The starting point is "Start."
 The first decision point is based on the weather forecast: "Sunny" or
"Rainy."
 If it's sunny, the next decision point is whether you have completed
all your pending tasks: "Tasks Done" or "Tasks Not Done."
 If your tasks are done, the final decision is to "Go on Trip."
 If your tasks are not done, the final decision is to "Stay at Home."
 If it's rainy, regardless of your pending tasks, the final decision is to
"Stay at Home."
Let's assume the following probabilities:
 P(Sunny) = 0.6 (60% chance of sunny weather)
 P(Rainy) = 0.4 (40% chance of rainy weather)
 P(Tasks Done | Sunny) = 0.8 (80% chance of completing tasks when
it's sunny)
 P(Tasks Not Done | Sunny) = 0.2 (20% chance of not completing tasks
when it's sunny)
Now, let's calculate the expected outcomes based on the decision tree:
Expected Outcome of Going on a Trip:
P(Go on Trip | Tasks Done, Sunny) = 1 (100% chance of going on a trip)
P(Go on Trip | Tasks Not Done, Sunny) = 0 (0% chance of going on a trip)
Page 46 of 74
Expected Outcome of Staying at Home:
P(Stay at Home | Rainy) = 1 (100% chance of staying at home)
Now, let's calculate the overall expected outcome:
Expected Outcome of Going on a Trip:
P(Go on Trip) = P(Sunny) * [P(Go on Trip | Tasks Done, Sunny) * P(Tasks
Done | Sunny) + P(Go on Trip | Tasks Not Done, Sunny) * P(Tasks Not Done
| Sunny)]
= 0.6 * [1 * 0.8 + 0 * 0.2]
= 0.6 * 0.8
= 0.48
Expected Outcome of Staying at Home:
P(Stay at Home) = P(Rainy) * P(Stay at Home | Rainy)
= 0.4 * 1
= 0.4
Therefore, the overall expected outcome of going on a trip is 0.48, and the
overall expected outcome of staying at home is 0.4.
Based on this analysis, it's better to go on a trip as it has a higher expected
outcome.
5.4 Bayes' Rule and Decision Trees
The general form of Bayes' rule, also known as Bayes' theorem, is as follows:
P(A | B) = (P(B | A) * P(A)) / P(B)
In this formula:
 P(A | B) represents the conditional probability of event A occurring
given that event B has occurred.
 P(B | A) represents the conditional probability of event B occurring
given that event A has occurred.
 P(A) is the prior probability of event A occurring.
 P(B) is the probability of event B occurring.
Bayes' rule allows us to update our belief about the probability of an event
A occurring based on new evidence provided by event B.
An expanded form of Bayes' rule, which includes additional terms to

consider multiple events and evidence:
P(A | B1, B2, ..., Bn) = (P(B1, B2, ..., Bn | A) * P(A)) / P(B1, B2, ..., Bn)
Page 47 of 74
In this expanded form:
 P(A | B1, B2, ..., Bn) represents the conditional probability of event A
occurring given that events B1, B2, ..., Bn have occurred.
 P(B1, B2, ..., Bn | A) represents the conditional probability of events
B1, B2, ..., Bn occurring given that event A has occurred.
 P(A) is the prior probability of event A occurring.
 P(B1, B2, ..., Bn) is the probability of events B1, B2, ..., Bn occurring.
This expanded form allows us to consider the joint probabilities of multiple
events (B1, B2, ..., Bn) and their relationship to event A. By incorporating
multiple evidence events, we can further refine our estimation of the
conditional probability of event A.
It's important to note that the joint probability P(B1, B2, ..., Bn | A) can be
calculated using the product rule of probability:
P(B1, B2, ..., Bn | A) = P(B1 | A) * P(B2 | A, B1) * P(B3 | A, B1, B2) * ... * P(Bn | A,
B1, B2, ..., Bn-1)
Each term in the product represents the conditional probability of an
evidence event given all the previous evidence events and event A.
The denominator P(B1, B2, ..., Bn) can be calculated using the law of total
probability, which accounts for all possible combinations of events:
P(B1, B2, ..., Bn) = P(B1, B2, ..., Bn | A) * P(A) + P(B1, B2, ..., Bn | A') * P(A')
Here, A' represents the complement of event A (i.e., the event A not
occurring).
By incorporating multiple evidence events and using the expanded form of
Bayes' rule, we can make more sophisticated probabilistic inferences and
update our beliefs based on complex combinations of evidence.
Bayes' Rule isa fundamental principle in probability theory that allows the
updating of probabilities based on new information. It is particularly useful
in decision-making under uncertainty when new evidence becomes
available. Bayes' Rule can be combined with decision trees to incorporate
new information and revise probabilities at each chance node. The steps to
incorporate Bayes' Rule into decision trees are as follows:
1. Construct the initial decision tree: Begin by constructing the decision
tree based on the available information and probabilities.
Page 48 of 74
2. Assign prior probabilities: Assign initial probabilities to the chance
nodes based on the available information before any new evidence is
considered.
3. Incorporate new evidence: When new evidence becomes available,
revise the probabilities at the chance nodes using Bayes' Rule. Bayes'
Rule states that the posterior probability of an event given new
evidence is equal to the prior probability of the event multiplied by
the probability of the evidence given the event, divided by the
probability of the evidence.
4. Update the decision tree: Update the decision tree by revising the
probabilities at the chance nodes based on the new evidence. This
involves adjusting the probabilities along the branches that originate
from the chance nodes.
5. Recalculate expected utilities: Recalculate the expected utilities or
values at the payoff nodes of the decision tree based on the revised
probabilities. This involves multiplying the revised probabilities by the
corresponding utilities or values and summing them up.
6. Analyze and make decisions: Analyze the updated decision tree to
determine the optimal decision path or alternative based on the
revised probabilities and expected utilities. The decision with the
highest expected utility or value is typically selected.
By incorporating Bayes' Rule into decision trees, decision makers can
update their probabilities and make more informed decisions as new
evidence becomes available. This allows for a dynamic and iterative
decision-making process that adapts to changing circumstances and
uncertainties.
Example
In a certain population, a rare disease affects 1 in every 10,000 individuals.
There is a diagnostic test available for this disease, which has the following
characteristics:
 The test correctly identifies a positive case 99% of the time
(sensitivity).
 The test correctly identifies a negative case 98% of the time
(specificity).
Now, suppose a randomly selected individual takes the test and receives a
positive result. We want to calculate the probability that this individual
actually has the disease.
Page 49 of 74
Solution
To solve this problem using Bayes' rule, we'll define the following events:
A: The individual has the disease.
B: The individual tests positive.
We want to calculate P(A | B), which represents the probability that the
individual has the disease given a positive test result.
Step 1: Determine the prior probabilities.
P(A) = Probability of having the disease = 1/10,000 = 0.0001
P(A') = Probability of not having the disease = 1 - P(A) = 0.9999
Step 2: Calculate the conditional probabilities.
P(B | A) = Probability of testing positive given the individual has the disease
= 0.99 (sensitivity)
P(B | A') = Probability of testing positive given the individual does not have
the disease = 1 - Specificity = 1 - 0.98 = 0.02
Step 3: Apply Bayes' rule.
P(A | B) = (P(A) * P(B | A)) / (P(A) * P(B | A) + P(A') * P(B | A'))
P(A | B) = (0.0001 * 0.99) / (0.0001 * 0.99 + 0.9999 * 0.02)
P(A | B) = 0.0099 / (0.0099 + 0.019998)
P(A | B) = 0.0099 / 0.029898
P(A | B) ≈ 0.3304
Therefore, the probability that the individual actually has the disease, given
a positive test result, is approximately 0.3304 or 33.04%.
-----------------------------------------------------------------------------------------------------
6.1 Deterministic Economic Order Quantity (EOQ) Model

The deterministic Economic Order Quantity (EOQ) model is an inventory
control model used to determine the optimal order quantity that minimizes
total inventory costs. It assumes that demand for the product is constant
and known with certainty, and there are no stockouts or shortages. The
model takes into account three main components:
1. Demand Rate (D): The demand rate represents the average number
of units sold or used per unit of time. It is assumed to be constant
throughout the planning period.
2. Ordering Cost (S): The ordering cost includes expenses related to
placing an order, such as administrative costs, transportation costs,
and order processing costs. It is assumed to be fixed per order.
3. Holding Cost (H): The holding cost represents the expenses
associated with holding inventory, such as storage costs, insurance,
Page 50 of 74
and opportunity cost of capital. It is assumed to be constant per unit
of inventory held over a given period.
The EOQ formula calculates the optimal order quantity (Q) as follows:
Q = √((2DS) / H)
Where:
Q = Optimal order quantity
D = Demand rate
S = Ordering cost per order
H = Holding cost per unit of inventory
Once the optimal order quantity is determined, the reorder point (ROP) can
be calculated as the demand rate multiplied by the lead time (L), where the
lead time is the time between placing an order and receiving the inventory.
ROP = D * L
By using the EOQ model, a company can determine the order quantity that
minimizes the total costs associated with inventory, finding the balance
between ordering costs and holding costs.
Worked Example for Deterministic EOQ Model:
Let's consider a company that sells widgets. The demand rate for widgets is
100 units per week, the ordering cost is $50 per order, and the holding cost
is $2 per unit per week. We can calculate the optimal order quantity using
the EOQ formula.
D = 100 units/week
S = $50/order
H = $2/unit/week
Q = √((2 * 100 * $50) / $2)
Q = √(10000 / $2)
Q = √5000
Q ≈ 70.71 units
The optimal order quantity for widgets is approximately 70.71 units per
order. This means that the company should place an order for 70.71 units
whenever the inventory level reaches the reorder point.
Example
A company sells a product with a constant demand rate of 500 units per
month. The ordering cost is $200 per order, and the holding cost is $2 per
unit per month. The company wants to determine the optimal order
quantity (EOQ) that minimizes total inventory costs.
Page 51 of 74
Solution
Step 1: Calculate the annual demand (D).
Annual Demand (D) = Demand Rate * 12 months
D = 500 units/month * 12 months
D = 6000 units
Step 2: Calculate the EOQ using the formula.
EOQ = √((2 * D * S) / H)
EOQ = √((2 * 6000 units * $200/order) / $2/unit/month)
EOQ = √((2400000 units * $200) / $2)
EOQ = √(480000000 units * $200 / $2)
EOQ = √(48000000000)
EOQ ≈ 6937 units
Therefore, the optimal order quantity (EOQ) for this scenario is
approximately 6937 units.
Step 3: Calculate the reorder point (ROP).
To calculate the reorder point, we need to consider the lead time (L). Let's
assume the lead time is 1 month.
Reorder Point (ROP) = Demand Rate * Lead Time
ROP = 500 units/month * 1 month
ROP = 500 units
Therefore, the reorder point (ROP) is 500 units.
By ordering the EOQ of approximately 6937 units whenever the inventory
level reaches the reorder point of 500 units, the company can minimize
total inventory costs by balancing ordering costs and holding costs.
6.2 Probabilistic Inventory Models

Probabilistic inventory models are used when demand for a product is
uncertain and follows a probability distribution. These models consider the
variability in demand and allow for a more realistic analysis of inventory
control. Two commonly used probabilistic inventory models are:
1. Periodic Review Model (s, S): In this model, inventory is reviewed at
fixed intervals, and orders are placed to bring the inventory level back
to a predetermined level (S) known as the order-up-to level. The
decision to place an order is based on the current inventory level (s)
and the reorder point (ROP). The ROP is calculated based on the
desired service level and the demand distribution.
Page 52 of 74
2. Continuous Review Model (Q, R): In this model, inventory is
continuously monitored, and orders are placed whenever the
inventory level reaches a reorder point (R). The order quantity (Q) is
predetermined and remains constant.
These models consider the trade-off between the costs associated with
holding excess inventory (holding costs) and the costs associated with
stockouts (shortage costs). By determining the optimal levels for reorder
points, order quantities, and safety stock, probabilistic inventory models
help minimize costs while ensuring an acceptable level of customer service.
Various probability distributions can be used to model demand, such as the
normal distribution, Poisson distribution, or exponential distribution. The
choice of distribution depends on the characteristics of the demand and
historical data.
To optimize inventory control in probabilistic models, additional metrics are
considered, including service level, fill rate, and safety stock. These metrics
help quantify the level of risk and customer service provided by the
inventory management strategy.
Probabilistic inventory models provide a more realistic approach to
inventory control by considering demand uncertainty and allowing for a
more accurate assessment of inventory costs and customer service levels.
Example
A company sells a product with an average demand rate of 200 units per
week. The standard deviation of demand is 50 units per week. The lead time
between placing an order and receiving the inventory is fixed at 2 weeks.
The company wants to determine the optimal reorder point (s) and order-
up-to level (S) to achieve a service level of 95%.
Solution
Step 1: Calculate the average demand during the lead time (L).
Average Lead Time Demand (L) = Average Demand Rate * Lead Time
L = 200 units/week * 2 weeks
L = 400 units
Step 2: Determine the safety stock (SS) based on the desired service level.
To calculate the safety stock, we need to find the z-value corresponding to
the desired service level. For a 95% service level, the z-value is
approximately 1.645 (obtained from a standard normal distribution table).
Safety Stock (SS) = Z-value * Standard Deviation * Square Root of Lead
Time
Page 53 of 74
SS = 1.645 * 50 units/week * √2 weeks
SS ≈ 115.9 units
Step 3: Calculate the reorder point (s) and order-up-to level (S).
Reorder Point (s) = Average Lead Time Demand + Safety Stock
s = 400 units + 115.9 units
s ≈ 515.9 units
Order-up-to Level (S) = Reorder Point + Average Demand during the
review period
S = s + Average Demand Rate * Review Period
Assume the review period is 1 week.
S = 515.9 units + 200 units/week * 1 week
S = 715.9 units
Therefore, in this scenario, the optimal reorder point (s) is approximately
515.9 units, and the order-up-to level (S) is approximately 715.9 units to
achieve a service level of 95%.
By periodically reviewing the inventory and placing orders to replenish
stock when it reaches the reorder point, the company can maintain a
desired service level while minimizing inventory costs.
-----------------------------------------------------------------------------------------------------
A1. (a) Customers arrive at a single server queueing system in accordance with a Poisson
system at rate 10 per hour. If the sever works continuously, the number of customers that can
be served in an hour has a Poisson distribution with mean 15.
Determine the proportion of time when noone is waiting to be served.
To determine the proportion of time when no one is waiting to be served,

we need to calculate the probability that the server is idle. In other words,
we need to find the probability that there are no customers in the system.
In a single server queueing system with a Poisson arrival rate of 10 per hour
and a Poisson service rate of 15 per hour, we can model the system using
the M/M/1 queueing model.
In the M/M/1 queueing model, the probability of having zero customers in
the system (P_0) can be calculated using the following formula:
P_0 = 1 - (λ/μ)
where λ is the arrival rate and μ is the service rate.
In this case, λ = 10 per hour (Poisson arrival rate) and μ = 15 per hour
(Poisson service rate).
Substituting the values into the formula:
Page 54 of 74
P_0 = 1 - (10/15)
= 1 - 2/3
= 1/3
Therefore, the proportion of time when no one is waiting to be served is
1/3, or approximately 0.3333 (33.33%).
Additional information for better understanding
The Poisson M/M/1 queueing model is a specific type of queueing model

that assumes a Poisson arrival process, exponential service times, and a
single server. It is commonly used to analyze and study the behavior of a
single server queueing system.
Here's a breakdown of the components in the M/M/1 queueing model:
1. Arrival Process: The arrival process follows a Poisson distribution,
where customers arrive independently and randomly over time. The
arrival rate is denoted by λ (lambda) and represents the average
number of arrivals per unit of time.
2. Service Time: The service time for each customer is exponentially
distributed with a mean of 1/μ, where μ (mu) is the average service
rate. The service rate represents the average number of customers
that can be served per unit of time.
3. Single Server: There is only one server available to serve customers.
The server works continuously without any breaks.
In the M/M/1 queueing model, the key assumptions include:
 Exponential inter-arrival times: The time between two consecutive
customer arrivals follows an exponential distribution.
 Exponential service times: The service time for each customer follows
an exponential distribution.
 First-come, first-served (FCFS) discipline: Customers are served in the
order they arrive.
 Infinite queue capacity: There is no limit to the number of customers
that can wait in the queue.
By analyzing the M/M/1 queueing model, various performance metrics can
be calculated, such as the average number of customers in the system, the
average waiting time, and the probability of different queueing scenarios
(e.g., no one in the system, specific number of customers in the system,
Page 55 of 74
etc.). These metrics help in understanding and optimizing the performance
of the queueing system.
A1 (b) Consider an item that is used at the rate of 5 units per day. The holding cost per
day is $1. The set up cost is $10. The unit purchase price is c1 = $2. If shortages
are not allowed, Determine the optimal order quantity.
we can use the Economic Order Quantity (EOQ) model. The EOQ model
helps to find the order quantity that minimizes the total cost of inventory,
considering the holding cost, setup cost, and purchase price.
In this case, we have the following information:
 Demand rate (D): 5 units per day
 Holding cost per day (H): $1
 Setup cost (S): $10
 Unit purchase price (c1): $2
The EOQ formula is given by:
EOQ = sqrt((2 * D * S) / H)
Substituting the values into the formula:
EOQ = sqrt((2 * 5 * 10) / 1)
= sqrt(100)
= 10
Therefore, the optimal order quantity is 10 units.
A2. (a) Define the following terms:

(i) maximax decision criterion,
(ii) minimax decision criterion, and
(iii) Bayes decision criterion
Page 56 of 74
(i) Maximax Decision Criterion:
The maximax decision criterion, also known as the optimistic criterion, is a
decision-making approach that aims to maximize the maximum possible
payoff or outcome. It assumes an optimistic perspective, where the
decision-maker seeks the best possible outcome under favorable
conditions. In other words, the maximax criterion selects the alternative
with the highest potential payoff among all possible outcomes.
(ii) Minimax Decision Criterion:
The minimax decision criterion, also known as the pessimistic criterion, is a
decision-making approach that focuses on minimizing the maximum
possible loss or regret. It assumes a pessimistic viewpoint, where the
decision-maker seeks to minimize the potential loss or regret under
unfavorable conditions. The minimax criterion selects the alternative with
the lowest maximum possible loss or regret among all possible outcomes.
(iii) Bayes Decision Criterion:
The Bayes decision criterion, also known as the Bayesian decision criterion,
is a decision-making approach that incorporates both prior probabilities
and the associated costs or payoffs of different decisions. It takes into
account the uncertainty of the decision-making environment by
considering the probabilities of different events or states of nature. The
Bayes criterion calculates the expected value or expected payoff for each
decision alternative, considering the probabilities and associated payoffs,
and selects the alternative with the highest expected value.
The maximax and minimax criteria are deterministic decision criteria,
focusing on extreme outcomes, while the Bayes criterion incorporates
probabilistic information to make decisions based on expected values.
A2 (b) A company is considering developing and marketing a new product. It is estimated to

be twice as likely that the product would prove to be successful as
unsuccessful. If it were successful, the expected profit would be $1 500 000. If
it is unsuccessful the expected loss is $1 800 000. A marketing survey that costs
$300 000 can be conducted to predict whether product would be successful. Past
experience with such surveys shows that successful products have been predicted
Page 57 of 74
to be successful 80% of the time whereas unsuccessful product have been predicted
to be unsuccessful 70% of the time
(i) Using revenue minus costs as payoff construct the decision tree.
(ii) Which action maximizes the expected payoff?
(i) To construct the decision tree, we will consider the different actions and
their associated probabilities and payoffs:
1. Action: Conduct Marketing Survey
 Cost: $300,000
 Probability of success prediction given a successful product:
0.80
 Probability of success prediction given an unsuccessful product:
0.30
 Probability of unsuccessful prediction given a successful
product: 0.20
 Probability of unsuccessful prediction given an unsuccessful
product: 0.70
2. Action: Do Not Conduct Marketing Survey
 No cost
Now, let's construct the decision tree:
Copy
[Conduct Marketing Survey]
/ \
[Successful Prediction] [Unsuccessful Prediction]
/ \ / \
[$1,500,000] [$1,500,000] [-$1,800,000] [-$1,800,000]
OR
[Do Not Conduct Marketing Survey]

/
[$1,500,000]
(ii) To determine the action that maximizes the expected payoff, we need to
calculate the expected payoffs for each action and select the one with the
highest expected value.
Expected Payoff for Conduct Marketing Survey:
= (0.80 * $1,500,000) + (0.20 * (-$1,800,000))
Page 58 of 74
= $1,200,000 - $360,000
= $840,000
Expected Payoff for Do Not Conduct Marketing Survey:
= $1,500,000
Comparing the expected payoffs, we can see that conducting the marketing
survey has an expected payoff of $840,000, while not conducting the survey
has an expected payoff of $1,500,000. Therefore, the action that maximizes
the expected payoff is to "Do Not Conduct Marketing Survey."
A3. An insurance company is introducing two new product lines: special risk insurance
and mortgages. The expected profit is 5 per unit on special risk insurance and 2 per
unit on mortgages. Management wishes to establish sales quotas for the new product
lines to maximize total expected profit. The work requirements are as follows:
W ork Hours per Unit
Department Special Risk Mortgage W ork Hours Available
Underwriting 3 2 2400
Administration 0 1 800
Claims 2 0 1200
(a) Formulate the linear programming problem. [3]
(b) use the graphical method to determine the optimal daily production lines.
(c) Use the simplex method to determine the optimal daily production of each model. [8
Let's define the decision variables as follows:

x = Number of units of special risk insurance to be sold
y = Number of units of mortgages to be sold
Objective function:
Maximize Z = 5x + 2y (Total expected profit)
Constraints:
Underwriting constraint: 3x + 2y ≤ 2400 (Work hours available in
underwriting department)
Administration constraint: y ≤ 800 (Work hours available in administration
department)
Claims constraint: 2x ≤ 1200 (Work hours available in claims department)
Non-negativity constraint: x ≥ 0, y ≥ 0 (Cannot have negative units)
Page 59 of 74
Therefore, the linear programming problem can be formulated as:
Maximize Z = 5x + 2y
subject to:
3x + 2y ≤ 2400
y ≤ 800
2x ≤ 1200
x ≥ 0, y ≥ 0
(b) To solve the linear programming problem graphically, we plot the
feasible region determined by the constraints and then identify the corner
point that maximizes the objective function.
(c) The simplex method is an iterative algorithm used to solve linear
programming problems. It involves constructing a simplex tableau and
performing iterations until an optimal solution is reached.
A4. (a) Describe when transportation problems are relevant in operations management.
(b) Perform two iteration of Vogel’s approximation method to solve the following
transportation problem.
Supplier Destination Supply
12345
1246575
Page 60 of 74
2763M46
3875256
4000004
Demand 4 4 2 5 5
(c) Give an example of an assignment problem.
(a) Transportation problems are relevant in operations management when

there is a need to allocate resources from multiple suppliers to multiple
destinations in an efficient and cost-effective manner. These problems
commonly arise in industries such as logistics, supply chain management,
distribution, and transportation. The goal is to minimize transportation
costs while meeting the demand requirements of the destinations.
(b) To solve the given transportation problem using Vogel's approximation
method, we follow these steps:
Iteration 1:
1. Calculate the penalty for each row by subtracting the two smallest
costs in that row.
Row 1: 4 - 2 = 2
Row 2: 6 - 3 = 3
Row 3: 7 - 2 = 5
Row 4: 0 - 0 = 0
2. Calculate the penalty for each column by subtracting the two smallest
costs in that column.
Column 1: 7 - 6 = 1
Column 2: 7 - 4 = 3
Column 3: 5 - 3 = 2
Column 4: 6 - 5 = 1
Column 5: 5 - 4 = 1
Page 61 of 74
3. Select the cell with the highest penalty (5 in Row 3) and allocate as
much as possible (2) to that cell.
Supplier 3 -> Destination 2: Allocate 2
4. Update the supply and demand values:
Supplier 3: Supply - 2 = 6
Destination 2: Demand - 2 = 2
Iteration 2:
1. Recalculate the penalties for the updated supply and demand values:
Row 1: 4 - 2 = 2
Row 2: 6 - 3 = 3
Row 4: 0 - 0 = 0
Column 1: 7 - 6 = 1
Column 3: 5 - 3 = 2
Column 4: 6 - 5 = 1
Column 5: 5 - 4 = 1
2. Select the cell with the highest penalty (3 in Row 2) and allocate as
much as possible (3) to that cell.
Supplier 2 -> Destination 3: Allocate 3
3. Update the supply and demand values:
Supplier 2: Supply - 3 = 4
Destination 3: Demand - 3 = 2
The resulting allocations are as follows:
Supplier 1 -> Destination 2: 2
(c) An example of an assignment problem is the assignment of workers to
tasks or jobs. Let's consider a scenario where a company needs to assign
four workers (W1, W2, W3, W4) to four tasks (T1, T2, T3, T4). Each worker
has a different skill level, and each task requires a specific skill level. The
goal is to find the optimal assignment that minimizes the total skill
mismatch or maximizes the total productivity.
Here's an example of the skill level matrix:
Copy
T1 T2 T3 T4
W1 5 9 8 7
W2 6 7 9 5
W3 8 6 7 4
W4 7 5 6 8
Page 62 of 74
The task is to assign each worker to a task in a way that maximizes the total
skill level or minimizes the total skill mismatch. This can be formulated as an
assignment problem, and various methods like the Hungarian algorithm
can be used to find the optimal assignment.
----------------------------------------------------------------------------------------------------------------
----------------------------------------------------------------------------------------------------------------
----------------------------------------------------------------------------------------------------------
A1. Define the following terms:
(a) statistic
(b) confidence interval
(c) level of significance
(d) Type I error
(e) Type II error
(a) Statistic: In statistics, a statistic refers to a numerical value or measure

calculated from a sample of data. It provides information or summarizes
certain aspects of the data. Statistics are commonly used to make
inferences or draw conclusions about a larger population based on the
information obtained from a smaller sample.
(b) Confidence Interval: A confidence interval is a range of values
constructed around a sample statistic to estimate an unknown population
parameter. It provides a range of plausible values within which the true
parameter value is likely to fall. The confidence level associated with the
interval represents the probability that the interval contains the true
parameter value.
(c) Level of Significance: The level of significance, often denoted by the
symbol α (alpha), is the predetermined threshold or probability used in
hypothesis testing to determine whether to reject the null hypothesis. It
represents the probability of making a Type I error (incorrectly rejecting a
true null hypothesis). Commonly used levels of significance are 0.05 (5%)
and 0.01 (1%).
(d) Type I Error: In hypothesis testing, a Type I error occurs when the null
hypothesis is incorrectly rejected, even though it is true. It represents a false
positive or a false alarm. The probability of committing a Type I error is
Page 63 of 74
equal to the level of significance (α) chosen for the test. It is also referred to
as a "false positive."
(e) Type II Error: In hypothesis testing, a Type II error occurs when the null
hypothesis is failed to be rejected, even though it is false. It represents a
failure to detect a true alternative hypothesis. The probability of committing
a Type II error is denoted by β (beta) and is related to the power of a
statistical test (1 - β). Type II error is also called a "false negative."
A2. (a) The standard deviation of an outlets daily sales is $810. The outlet is to be closed
down unless daily sales become more than $2000. A random sample of sales on 9
days is : 2000, 1800, 1900, 3000, 1500, 2000, 1900, 1000, 2000. Using a t-test and
= 0 : 10, test H0 : µ = $2000 against H1 : 6= $2000.
(b) A manufacturer has advertised washing powder for sale in packages weighing 30kg
on average. The manufacturer is worried that if he sells packages weighing less
than 30kg on average he will lose contacts; and if he sells more than 30kg on
average he will waste the product. A sample of 75 packages has produced an
average weight of 29kg. Assuming that the packages have a standard deviation
of 1.5kg.
(i) Find the 95% confidence interval for the mean weight of packages.
(ii) State the null and the alternative hypothesis.
(iii) Determine if the sample mean weight of 29kg should be acceptable to the
manufacturer.
(a) To test the hypothesis H0: µ = $2000 against H1: µ ≠ $2000, we can use
a t-test since we have a small sample size (n = 9) and the population
standard deviation is unknown. Here are the steps to perform the t-test:
Step 1: Calculate the sample mean (x̄ ) and sample standard deviation (s)
from the given data:
Sample mean (x̄ ) = (2000 + 1800 + 1900 + 3000 + 1500 + 2000 + 1900 +
1000 + 2000) / 9 = 1888.89
Sample standard deviation (s) = sqrt(((2000 - 1888.89)^2 + (1800 -
1888.89)^2 + ... + (2000 - 1888.89)^2) / (9 - 1))
Step 2: Calculate the test statistic (t-value) using the formula:
t = (x̄ - µ) / (s / sqrt(n))
Page 64 of 74
Step 3: Determine the critical value for the desired level of significance (α)
and degrees of freedom (df = n - 1). Since the alternative hypothesis is two-
tailed (µ ≠ $2000) and α = 0.10, we divide the significance level by 2 to get
α/2 = 0.05 and find the critical value from the t-distribution table with df =
8 and α/2 = 0.05.
Step 4: Compare the absolute value of the test statistic (|t|) with the critical
value to make a decision. If |t| > critical value, we reject the null hypothesis.
Otherwise, we fail to reject the null hypothesis.
Let's calculate the values:
Step 1:
x̄ = 1888.89
s = sqrt(((2000 - 1888.89)^2 + (1800 - 1888.89)^2 + ... + (2000 -
1888.89)^2) / (9 - 1))
Step 2:
t = (x̄ - µ) / (s / sqrt(n))
Step 3:
Critical value for df = 8 and α/2 = 0.05 from the t-distribution table: ±2.306
Step 4:
|t| > critical value? If yes, reject H0. If no, fail to reject H0.
Now, let's perform the calculations:
Step 1:
x̄ = 1888.89
s = sqrt(((2000 - 1888.89)^2 + (1800 - 1888.89)^2 + ... + (2000 -
1888.89)^2) / (9 - 1))
s ≈ 543.43
Step 2:
t = (x̄ - µ) / (s / sqrt(n))
t = (1888.89 - 2000) / (543.43 / sqrt(9))
t ≈ -1.99
Step 3:
Critical value for df = 8 and α/2 = 0.05: ±2.306
Step 4:
|t| = 1.99 < 2.306
Since |t| < critical value, we fail to reject the null hypothesis H0: µ = $2000.
Therefore, based on the given data, we do not have sufficient evidence to
conclude that the mean daily sales are significantly different from $2000 at
a significance level of 0.10.
Page 65 of 74
(b) (i) To find the 95% confidence interval for the mean weight of packages,
we can use the formula:
Confidence Interval = sample mean ± (critical value * standard error)
The critical value depends on the desired confidence level and the sample
size. For a 95% confidence level, the critical value can be obtained from the
standard normal distribution table, which corresponds to a two-tailed test
with α = 0.05. The standard error can be calculated by dividing the standard
deviation by the square root of the sample size.
Let's calculate the confidence interval:
Sample mean (x̄ ) = 29kg
Standard deviation (σ) = 1.5kg
Sample size (n) = 75
Standard error = σ / sqrt(n) = 1.5 / sqrt(75) ≈ 0.172 kg
Critical value for α/2 = 0.05/2 = 0.025 from the standard normal
distribution table is approximately 1.96.
Confidence Interval = 29 ± (1.96 * 0.172)
Lower bound = 29 - (1.96 * 0.172) ≈ 28.663 kg
Upper bound = 29 + (1.96 * 0.172) ≈ 29.337 kg
The 95% confidence interval for the mean weight of packages is
approximately 28.663 kg to 29.337 kg.
(ii) Null Hypothesis (H0): The mean weight of packages is 30 kg.
Alternative Hypothesis (H1): The mean weight of packages is not 30 kg.
(iii) To determine if the sample mean weight of 29 kg should be acceptable
to the manufacturer, we need to check if the sample mean falls within the
95% confidence interval. If the sample mean falls within the confidence
interval, it suggests that the true population mean may be 30 kg, and the
sample is consistent with the manufacturer's requirement. However, if the
sample mean falls outside the confidence interval, it suggests that the true
population mean is likely different from 30 kg, and the sample may not be
acceptable to the manufacturer.
In this case, the sample mean weight of 29 kg falls within the 95%
confidence interval of approximately 28.663 kg to 29.337 kg. Therefore,
based on the given data, the sample mean weight of 29 kg should be
acceptable to the manufacturer.
A3. (a) Explain the difference between correlation and regression.

(b) A plant distills liquid air to produce oxygen, nitrogen, and argon. The percentage
Page 66 of 74
of impurity in the oxygen is thought to be linearly related to the amount of impurities in the
air as measured by the \pollution count" in parts per million(ppm).
A sample of plant operating data is shown below
Purity(y) Pollutioncount(x)
93.3 1.10
92.0 1.45
92.4 1.36
91.7 1.59
94.0 1.08
(i) Obtain the linear regression relationship using formulae.
(ii) Use SPSS to obtain the model.
(iii) Test the hypothesis that there is a relationship between pollution and purity.
(a) Correlation and regression are both statistical techniques used to

analyze the relationship between variables, but they serve different
purposes.
Correlation measures the strength and direction of the linear relationship
between two variables. It quantifies how closely the variables are related to
each other. Correlation coefficients range from -1 to +1, where -1 indicates
a perfect negative linear relationship, +1 indicates a perfect positive linear
relationship, and 0 indicates no linear relationship. Correlation does not
imply causation and only assesses the association between variables.
Regression analysis is used to model the relationship between a dependent
variable and one or more independent variables. It aims to find the best-
fitting line (or curve) that represents the relationship between the variables.
Regression provides insights into the nature of the relationship, as well as
predictions or estimates of the dependent variable based on the
independent variable(s). It can be used for prediction, forecasting, and
understanding the impact of independent variables on the dependent
variable.
(b) To obtain the linear regression relationship using formulae, we need to
calculate the slope and intercept of the regression line.
(i) Using the formulae:
Step 1: Calculate the means of purity (ȳ) and pollution count (x̄ ) from the
given data.
Page 67 of 74
Step 2: Calculate the deviations from the means: (x - x̄ ) and (y - ȳ).
Step 3: Calculate the product of the deviations: (x - x̄ )(y - ȳ) for each data
point.
Step 4: Calculate the sum of the product of deviations: Σ[(x - x̄ )(y - ȳ)].
Step 5: Calculate the sum of squared deviations of x: Σ(x - x̄ )^2.
Step 6: Calculate the slope (β1): β1 = Σ[(x - x̄ )(y - ȳ)] / Σ(x - x̄ )^2.
Step 7: Calculate the intercept (β0): β0 = ȳ - β1 * x̄ .
(ii) SPSS software can be used to obtain the linear regression model by
inputting the data and running the regression analysis. It provides the
regression coefficients, standard errors, p-values, and other relevant
statistics.
(iii) To test the hypothesis that there is a relationship between pollution and
purity, we can perform a hypothesis test on the slope (β1) of the regression
line. The null hypothesis (H0) states that there is no relationship (β1 = 0),
while the alternative hypothesis (H1) states that there is a relationship (β1 ≠
0).
The hypothesis test can be conducted by calculating the t-statistic using the
formula:
t = (β1 - 0) / SE(β1),
where SE(β1) is the standard error of the slope.
The t-statistic can be compared to the critical values of the t-distribution
with n - 2 degrees of freedom (n = sample size). If the t-statistic is
significant (p-value < α), we reject the null hypothesis and conclude that
there is a relationship between pollution and purity.
VOGEL WORKED EXAMPLE
Vogel’s Approximation Method Solved Example

Question:
Solve the given transportation problem using Vogel’s approximation method.
Page 68 of 74
Solution:
For the given cost matrix,
Total supply = 50 + 60 + 25 = 135
Total demand = 60 + 40 + 20 + 25 = 135
Thus, the given problem is balanced transportation problem.
Now, we can apply the Vogel’s approximation method to minimize the total cost of
transportation.
Step 1: Identify the least and second least cost in each row and column and then write the
corresponding absolute differences of these values. For example, in the first row, 2 and 3 are
the least and second least values, their absolute difference is 1.
These row and column differences are called penalties.
Step 2: Now, identify the maximum penalty and choose the least value in that corresponding
row or column. Then, assign the min(supply, demand).
Page 69 of 74
Here, the maximum penalty is 3 and the least value in the corresponding column is 2. For this
cell, min(supply, demand) = min(50, 40) = 40
Allocate 40 in that cell and strike the corresponding column since in this case demand will be
satisfied, i.e., 40 – 40 = 0.
Step 3: Now, find the absolute row and column differences for the remaining rows and
columns. Then repeat step 2.
Here, the maximum penalty is 3 and the least cost in that corresponding row is 3. Also, the
min(supply, demand) = min(10, 60) = 10
Thus, allocate 10 for that cell and write down the new supply and demand for the
corresponding row and column.
Supply = 10 – 10 = 0
Demand = 60 – 10 = 50
As supply is 0, strike the corresponding row.
Page 70 of 74
Step 4: Repeat the above step, i.e., step 3. This will give the below result.
In this step, the second column vanishes and the min(supply, demand) = min(25, 50) = 25 is
assigned for the cell with value 2.
Step 5: Again repeat step 3, as we did for the previous step.
Page 71 of 74
In this case, we got 7 as the maximum penalty and 7 as the least cost of the corresponding
column.
Step 6: Now, again repeat step 3 by calculating the absolute differences for the remaining
rows and columns.
Step 7: In the previous step, except for one cell, every row and column vanishes. Now,
allocate the remaining supply or demand value for that corresponding cell.
Page 72 of 74
Total cost = (10 × 3) + (25 × 7) + (25 × 2) + (40 × 2) + (20 × 2) + (15 × 3)
= 30 + 175 + 50 + 80 + 40 + 45
= 420
Vogel’s Approximation Method Problems
1. Consider the transportation problem given below. Solve this problem by Vogel’s
approximation method.
Destination
Origin Supply
D1 D2 D3 D4
O1 3 1 7 4 300
O2 2 6 5 9 400
O3 8 3 3 2 500
Demand 250 350 400 200
2. Find the solution for the following transportation problem using VAM.
To
From Supply
D1 D2 D3
A1 6 8 10 150
A2 7 11 11 175
A3 4 5 12 275
Demand 200 100 350
3. Use Vogel’s Approximation Method to find a basic feasible solution for the following.
Page 73 of 74
Destinations
Sources Supply
D1 D2 D3
S1 4 5 1 40
S2 3 4 3 60
S3 6 2 8 70
Demand 70 40 60
Page 74 of 74

Business Statistics and Management Science Notes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Business Statistics and Management Science Notes

Uploaded by

Copyright:

Available Formats

Statistical Inference: Hypothesis Testing

1.1 Testing of Hypothesis

1.4 Chi-square Test for Goodness of Fit

where O represents the observed frequency and E represents the

where O represents the observed frequency and E represents the

Correlation and Regression Analysis

2.2 Multiple Regression Analysis

To perform multiple regression analysis using SPSS, follow these steps:

1. Launch SPSS and open a new data file.

Interpreting the results:

1. Coefficients: The coefficient values represent the estimated effect of each

- β₀ (intercept): The estimated sales revenue when all independent variables

2. Standard Errors: These values provide an indication of the precision of the

4. R² and Adjusted R²: These measures indicate the proportion of variability in

3.3 What is Linear Programming?

3.5 The Simultaneous Equation Approach to a 2-Variable LP

3.7 Blending Problems

4.1 Formulating Transportation Problems

4.2 Assignment Problems

4.3 Transhipment Problems

5.1 Decision Criteria

2. Maximin Criterion: This criterion selects the alternative that maximizes

3. Maximizing Expected Monetary Value (EMV): This criterion calculates

To apply the Maximizing EMV criterion, we calculate the expected monetary

5. Minimax Regret Criterion: This criterion focuses on the regret

6. Hurwicz Criterion: This criterion involves a coefficient of optimism (α)

5.2 Utility Theory

Utility theory is a decision-making framework that incorporates the

5.3 Decision Trees

5.4 Bayes' Rule and Decision Trees

An expanded form of Bayes' rule, which includes additional terms to

6.1 Deterministic Economic Order Quantity (EOQ) Model

6.2 Probabilistic Inventory Models

To determine the proportion of time when no one is waiting to be served,

Additional information for better understanding

The Poisson M/M/1 queueing model is a specific type of queueing model

A2. (a) Define the following terms:

A2 (b) A company is considering developing and marketing a new product. It is estimated to

unsuccessful. If it were successful, the expected profit would be $1 500 000. If

to be unsuccessful 70% of the time

(ii) Which action maximizes the expected payoff?

[Do Not Conduct Marketing Survey]

W ork Hours per Unit

Department Special Risk Mortgage W ork Hours Available

(a) Formulate the linear programming problem. [3]

Let's define the decision variables as follows:

(c) Give an example of an assignment problem.

(a) Transportation problems are relevant in operations management when

(a) Statistic: In statistics, a statistic refers to a numerical value or measure

A3. (a) Explain the difference between correlation and regression.

(a) Correlation and regression are both statistical techniques used to

VOGEL WORKED EXAMPLE

Vogel’s Approximation Method Solved Example

Solve the given transportation problem using Vogel’s approximation method.

For the given cost matrix,

Total supply = 50 + 60 + 25 = 135

Total demand = 60 + 40 + 20 + 25 = 135

Thus, the given problem is balanced transportation problem.

These row and column differences are called penalties.

As supply is 0, strike the corresponding row.

Step 5: Again repeat step 3, as we did for the previous step.

Vogel’s Approximation Method Problems

You might also like