You are on page 1of 72

1

Indian Institute of €Management Bangalore


Decision Science II
Old Exams

 Please do not print this unless you really need to.


Midterm:
Question 1 (15 points)

Goodhealth Pharmacy runs a chain of stores across 50 regions in South and West India.
The company faces strong competition and must budget for promotional expenditure
towards advertisements and discounts across these regions. Data on Promote and Sales,
as described below, was collected from these regions to understand the impact of
promotion expenditure on sales:

Promote: Promotional expenditure as a percentage of those of the leading competitor


Sales: Sales figures as a percentage of those of the leading competitor

Table 1.1 Descriptive Statistics


N Mean Std. Deviation
Promote 50 97.88 8.736
Sales 50 99.74 9.895

A simple linear regression model was developed between Sales as the dependent variable
and Promote as the independent variable. The regression output using Excel is given
below in Table 1.2.
Table 1.2 Regression Output

Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations

ANOVA
  df SS MS F Significance F
Regression 39.736 8.6011E-08
Residual
Total 4797.62
2

Coefficient P-
Standard Error t Stat Lower 95% Upper 95%
s value
Intercept 11.8826 2.1146 0.0397 1.2349 49.0180
Promote

Hint: Use given values of the F-stat and SST as a starting point to calculate missing entries in
Table 1.2, required for answering the questions below.

Question 1.1
Using the Simple Linear Regression model fitted to the data, predict the value of Sales
when Promote has a value of 80.

Question 1.2
Is there a statistically significant relationship between Sales and Promote at a 2%
significance level? State the hypothesis clearly.

Question 1.3
What is the proportion of the variation in Sales that is explained by Promote?

Question 1.4
What is the lower limit of the 95% prediction interval for a specific Sales value when
Promote takes a value of 80?

Question 1.5
Goodhealth Pharmacy believes that if their promotional expenditure (as a percentage of
those of the leading competitor) is increased by 1%, their sales (as a percentage of those
of the leading competitor) will go up by at least 0.6%. Conduct an appropriate test to
check if this hypothesis is correct at a 5% significance level. State the hypothesis clearly.
3

The Normal P-P plot and residual plot are shown in Figures 1.1 and 1.2.

Figure 1.1 Normal P-P Plot

Figure 1.2 Residual Plot

Question 1.6
Use the plots above to conclude whether the various assumptions made for linear
regression are satisfied or not. Explain clearly.
4

Question 1.7
Do you recommend using the current model for prediction? Clearly explain why or why
not?

Question 2 (20 points)

A survey of 113 hospitals in the United States was conducted by the Department of
Health Services (DHS) to measure the risk of infection for patients while they are
admitted in a hospital. The variables recorded in the study are given in the table below:

Table 2.1. Variable names, descriptions, and types


Variable Variable Description Variable Type
InfctRsk Risk of infection while admitted in a hospital Numerical
Stay Average number of days a patient stays in a hospital Numerical
Beds Number of in-patient beds in a hospital Numerical
X-Ray Average number of X-rays given to a patient * 10 Numerical
Nurses Number of nurses employed by a hospital Numerical
Age Age of a patient Numerical
Categorical
MedSchool Hospital attached to a teaching institution
(Yes = 1/No = 0)
Region Region where hospital is located Categorical (1, 2, 3, 4)

The (partially completed) pairwise correlation matrix for a subset of these variables is
first computed as shown below:

Table 2.2 Correlation matrix


  InfctRsk Stay Beds Xray Nurses
InfctRsk 1
Stay 0.533 1
Beds 0.360 0.409 1
Xray 0.453 0.046 1
Nurses 0.394 0.340 0.077 1

To begin with, a stepwise regression model was developed to predict infection risk as a
function of the predictor variables given in the correlation table above, and the output is
shown below:
5

Regression #1: Answer Questions 2.1 – 2.5 using this output

Model Summaryd
Adjusted R Std. Error of the
Model R R Square
Square Estimate
1
2
3 .645c .416 .400
a. Predictors: (Constant), ________
b. Predictors: (Constant), ________, Xray
c. Predictors: (Constant), ________, Xray, Nurses
d. Dependent Variable: InfctRsk

ANOVAa
Model Sum of Squares df Mean Square F Sig.
1 Regression .000
Residual
Total
2 Regression .000
Residual
Total
3 Regression 83.840 25.916 .000
Residual 117.540
Total
a. Dependent Variable: InfctRsk
b. Predictors: (Constant), ________
c. Predictors: (Constant), ________, Xray
d. Predictors: (Constant), ________, Xray, Nurses

Coefficientsa
Unstandardized Coefficients Standardized Coefficients
Model t Sig.
B Std. Error Beta
(Constant) .744 .554 1.344 .182
1
.374 .056 6.645 .000
(Constant) -.151 .585 -.257 .797
2 .058 .422 5.098 .000
Xray .006 .292 3.531 .001
(Constant) -.035 .561 -.063 .950
.230 .059 .328 3.894 .000
3
Xray .021 .005 .308 3.883 .000
Nurses .002 .001 .259 3.317 .001
a. Dependent Variable: InfctRsk
Question 2.1
6

What is the predictor variable used in Model 1?

Question 2.2
Complete all the entries in the ANOVA and Model Summary tables for Model 1.

Question 2.3
At the end of Step 1 in the stepwise regression, the partial correlation of the variable ‘X-
ray’ is 0.319. What proportion of the variability in ‘Infection Risk’ is explained by Model
2? What is the part correlation of the variable ‘X-ray’ in Model 2?

Question 2.4
Determine the coefficients of the ‘predictor variable used in Model 1’ and ‘X-ray’ in
Model 2. Based on this information, what can you conclude about the relationship
between ‘predictor variable in Model 1’ and ‘X-ray’?

Question 2.5
At the end of the stepwise regression procedure, it was observed that one of the candidate
predictor variables is excluded from the final model. Therefore, it was concluded that the
excluded variable is not correlated to ‘Infection Risk’. Do you agree with this
conclusion? Justify your answer.
7

Regression #2: Answer Questions 2.6 – 2.8 using this output

The Director of the DHS is interested in determining whether the location of the hospital
is a factor in explaining the risk of infections. A linear regression model was developed
by including the variable ‘Region’, and the SPSS output is given below:

Model Summaryb
Model R R Square Adjusted R Square Std. Error of the Estimate
1 .669a
.448 .416 1.0245
a. Predictors: (Constant), Region3, Nurses, Xray, Region2, Stay, Region1
b. Dependent Variable: InfctRsk

ANOVAa
Model Sum of Squares df Mean Square F Sig.
Regression 90.123 6 15.021 14.311 .000b
1 Residual 111.257 106 1.050
Total 201.380 112
a. Dependent Variable: InfctRsk
b. Predictors: (Constant), Region3, Nurses, Xray, Region2, Stay, Region1

Coefficientsa
Unstandardized Coefficients Standardized Coefficients
Model t Sig.
Beta
B Std. Error
1 (Constant) .171 .611 .280 .780
Stay .279 .065 .398 4.289 .000
Xray .021 .006 .298 3.733 .000
Nurses .002 .001 .248 3.210 .002
Region1 -.740 .367 -.239
Region2 -.640 .327 -.216
Region3 -.743 .315 -.261
a. Dependent Variable: InfctRsk

Question 2.6
Using the coefficients given in Regression #2, rank the different regions starting from
best to worst for the variable ‘Infection Risk’. (For now, ignore whether the ‘Region’
coefficients are significant or not.)
8

Question 2.7
Are the coefficients of the variables ‘Region 1’, ‘Region 2’, and ‘Region 3’ significant at
a 99% confidence level? Clearly state the null and alternate hypothesis and state your
conclusions.

Question 2.8
Suppose that Director of DHS claims that ‘Region’ does not have any impact on Infection
Risk. Perform an appropriate hypothesis test and verify this claim at a 95% confidence
level. Clearly show all your work.

Question 3 (15 points)

Elec-Mart is an on-line store that specializes in office supplies. To understand and predict
the behavior of on-line purchasing, a ‘partial’ snapshot of the transactions made by
customers is shown below. That is, each record represents data on a specific transaction
made by a customer.

CardType BuyCategory ItemsOrdered TotalValue


ElecMart High 4 $136.97
Other Medium 1 $25.55
ElecMart Medium 5 $113.95
Other Low 1 $6.82
ElecMart Medium 4 $147.32
Other Medium 5 $142.15
Other Low 1 $18.65
Other High 4 $178.34
Other Low 2 $25.83
Other Low 1 $18.13
ElecMart Medium 2 $54.52
Other Medium 2 $61.93
ElecMart High 3 $147.68

In the table above, TotalValue measures the total $ value of the items purchased, while
ItemsOrdered measures the number of items purchased, which varies from 1 to 10.
BuyCategory represents a categorization of that customer’s previous transaction
purchase value. High represents high purchase value, Medium - a medium purchase value
and Low – a low purchase value. Finally, CardType records the type of credit card used
for that transaction. ElecMart denotes use of the credit card issued by Elec-Mart, while
9

Other represents any other credit card. A random sample of 400 such transactions were
recorded.

To construct linear regression models the following indicator variables are defined:

Buy-Med = 1 if BuyCategory = Medium, 0 otherwise.


Buy-High = 1 if BuyCategory = High, 0 otherwise.
ElecMart-C = 1 if CardType = ElecMart, 0 otherwise.

In addition, the following additional variables are defined as well:

Buy-Med_Ord = Buy-Med*ItemsOrdered
Buy-High_Ord = Buy-High*ItemsOrdered.
ElecMart_Ord = ElecMart-C*ItemsOrdered.

The following outputs, in which TotalValue is the response variable, were generated
using this data:
Regression Output 1:

Regression Statistics
Multiple R 0.908
R Square 0.824
Adjusted R
0.823
Square
Standard Error 40.969
Observations 400

ANOVA
  df SS MS F Significance F
Regression 3 3118440.9 1039480.3 619.296 0
Residual 396 664680.94 1678.487
Total 399 3783121.9

Upper
  Coefficients Standard Error t Stat P-value Lower 95.0%
95.0%
Intercept 14.945 4.313 3.465 0.001 6.465 23.424
Buy-Med 14.674 5.154 2.847 0.005 4.541 24.807
Buy-High 77.043 6.174 12.48 0 64.906 89.18
ItemsOrdered 32.115 1.266 25.367 0 29.626 34.604
10

Regression Output 2:

Regression Statistics
Multiple R 0.925
R Square 0.855
Adjusted R
0.853
Square
Standard Error 37.274
Observations 400

ANOVA
  df SS MS F Significance F
Regression 4 3234326.3 808581.58 581.983 0
Residual 395 548795.56 1389.356
Total 399 3783121.9      

Coefficient
  Standard Error t Stat P-value
s
Intercept 51.001 5.056 10.088 0
Buy-High -10.894 10.41 -1.047 0.296
Buy-Med_Ord 9.708 1.6 6.067 0
Buy-High_Ord 26.06 2.706 9.63 0
ItemsOrdered 16.096 2.157 7.461 0

Regression Output 3:

Regression Statistics
Multiple R 0.873
R Square 0.763
Adjusted R
0.761
Square
Standard Error 47.56
Observations 400

ANOVA
Significance
  df SS MS F
F
Regression 2 2885119.6 1442559.8 637.745 0
Residual 397 898002.24 2261.97
Total 399 3783121.9

Coefficient
  Standard Error t Stat P-value
s
Intercept 8.782 4.726 1.858 0.064
11

ElecMart-C 28.485 5.87 4.853 0


ItemsOrdered 39.897 1.232 32.376 0

Regression Output 4:

Regression Statistics
Multiple R 0.876
R Square 0.768
Adjusted R
0.766
Square
Standard Error 47.09
Observations 400

ANOVA
  df SS MS F Significance F
Regression 3 2904993.9 968331.3 436.678 0
Residual 396 878127.99 2217.495
Total 399 3783121.9

Coefficient
  Standard Error t Stat P-value
s
Intercept 17.39 5.492 3.166 0.002
ElecMart-C -2.212 11.786 -0.188 0.851
ElecMart_Ord 7.606 2.541 2.994 0.003
ItemsOrdered 37.153 1.526 24.346 0

Using the information given in the regression tables, answer the following questions. For
each question give adequate explanation and support your answer with given
information precisely.

Question 3.1
Determine the best possible estimate of the average value of an additional item purchased
in a transaction by a repeat customer (a customer with a purchase history) chosen at
random.

Question 3.2
Determine the best possible estimate of the average value of the 1 st item purchased by a
customer chosen at random from those that use a credit card other than ElecMart credit
card.
12

Question 3.3
If a customer who purchased a medium amount during her last transaction purchases 7
items this time, determine the best possible point estimate of the total value of her
purchase.

Question 3.4
If a repeat customer who purchased a low amount during his last transaction purchases 2
items during this transaction, determine the best possible estimate of the probability that
his purchase value will be at least $80.

Question 3.5
Instead of Buy-Med, let Buy-Low be defined as Buy-Low = 1 if BuyCategory = Low, 0
otherwise. Further, define Buy-Low_Ord = Buy-Low*ItemsOrdered. Restate prediction
equation in Model 2 in terms ItemsOrdered, Buy-High, Buy-Low, Buy-High_Ord and
Buy-Low_Ord.
13

Question 1

It is believed that, caste plays an important role in Indian elections. Data is collected from 2019
elections from 37 parliamentary constituencies of Tamil Nadu for the political party called
A2ZMK (A to Z Munnetra Kazhagam) and number of people belonging A to Z (A2Z) caste in each
of the 37 constituencies. Note that, depending on the candidate, they also get large number of
votes from people belonging to other castes. Data description is provided in Table 1.1 and
Descriptive Statistics are shown in Table 1.2.

Table 1.1 Data Description

Variable Name Description


NVP Votes polled for A2ZMK in a
constituency
A2Z Number of voters belonging to A2Z
caste in a constituency
ln(NVP) = ln(NVP)
ln(A2Z) = ln(A2Z)

Table 1.2 Descriptive Statistics

N Mean Std. Deviation


A2Z 37 17536.000 11303.7003
NVP 37 94623.35 6924.458
LNA2Z 37 9.5855 .6298
LNNVP 37 11.4550 .0733
Valid N (listwise) 37

A simple linear regression model was developed between the variable NVP as dependent
variable and A2Z as independent variable. The regression outputs using SPSS are shown in
Tables 1.3 and 1.4.

Table 1.3 Model Summaryb

Model R R Square Adjusted R Std. Error of the


Square Estimate

1 4722.744

a. Predictors: (Constant), A2Z


b. Dependent Variable: NVP
Table 1.4 Coefficientsa

Model Unstandardized Coefficients Standardized t Sig.


Coefficients

B Std. Error Beta

1 (Constant) 86673.0292 1447.0391 59.897 .000


14

A2Z .070 .740 6.511

a. Dependent Variable: NVP

Question 1.1

Is there a statistically significant relationship between votes polled (NVP) for A2ZMK and the
number of voters belonging to A2Z caste in the constituency at a 1% significance level?

Question 1.2
A constituency has 50000 voters belonging the A2Z caste. Predict the votes polled for A2ZMK in
this constituency.

Question 1.3
What proportion of the variation in votes polled for A2ZMK is explained by the number of
people belonging the A2Z caste?

Question 1.4
The president of the A2ZMK believes that at least 40% of the people belonging to the A2Z caste
vote for them. Check whether the president’s claim is true at a 10% significance level. Clearly
write all the steps.

Question 1.5
Any party that gets more than 40% of the polled votes in a constituency is likely to win the
constituency. If a constituency has 4,00,000 polled votes, calculate the probability that A2ZMK
will win the election in that constituency. Assume that the number of voters belonging to A2Z
caste is 1,50,000.

Question 1.6
If a specific constituency has 50,000 people belonging to A2Z caste, what is the maximum votes
A2ZMK is likely to get at a 95% confidence level? Clearly write all the steps.

The normal P-P plot and residual plots are shown in Figures 1.1 and 1.2.
15

Figure 1.1 Normal P-P Plot

Figure 1.2 Residual plot

Question 1.7

Based on the P-P plot (Figure 1.1) and Residual Plot (1.2), comment on the validity of model
shown in Tables 1.3 and 1.4. Clearly identify any potential problem with the model.
16

A second model is developed between ln(NVP) and ln(A2Z). The model outputs are provided in
Tables 1.5 and 1.6 and Figures 1.3 and 1.4.

Table 1.5 Model Summaryb

Model R R Square Adjusted R Std. Error of the


Square Estimate
a
1 .754 .0488

a. Predictors: (Constant), LNA2Z


b. Dependent Variable: LNNVP

Table 1.6 Coefficientsa

Model Unstandardized Coefficients Standardized t Sig.


Coefficients

B Std. Error Beta

(Constant) 10.614 .124 85.500 .000


1
LNA2Z .088 .013

a. Dependent Variable: LNNVP

Question 1.8
What is the interpretation of the regression coefficient 0.088 in Table 1.6 with respect to the
votes polled?

Question 1.9
Coimbatore has 50,000 voters belonging A2Z caste, what is the number of the votes polled for
A2ZMK in Coimbatore using Model 2?

Question 2

Meddicorp sells medical supplies to hospitals, clinics and doctors’ offices. The company
currently markets in 3 regions: South, West and North. These regions are divided into many
smaller sales territories. Meddicorp’s sales manager is concerned about the effectiveness of a
new bonus program that provides bonuses to salespersons based on their performance. He
would like to know if the bonuses paid in 2018 were related to sales, taking the effect of
advertising into account.

Data on SALES (sales in thousands of dollars), ADV (advertising expenditure in hundreds of


dollars) and BONUS (bonuses paid in hundreds of dollars) for 60 sales territories has been
collected for 2018.

Descriptive Statistics

Mean Std. Deviation N

SALES 1285.842 222.265 60


BONUS 277.821 26.720 60
ADV 520.535 70.227 60
17

Correlation Matrix
  SALES ADV BONUS
SALES 1
ADV 0.891 1
BONUS 0.554 0.396 1

A multiple linear regression equation is developed using the Forward Method on SPSS and the
model summary is given below:

Model 1 and 2 Summaryc

Model R R Square Adjusted R Std. Error of the


Square Estimate

1 .891a .794 .790 101.8198


2

a. Predictors: (Constant), ADV c. Dependent Variable: SALES


b. Predictors: (Constant), ADV, BONUS

Model 1 and 2 Coefficientsa


Models Unstandardized Standardized t Sig. Correlations
Coefficients Coefficients

B Std. Beta Zero- Partial Part


Error order
1 (Constant) -181.89 99.130   -1.835 .072      

ADV 2.820 .189 .891 14.938 .000


2 (Constant) -576.70 129.433   -4.456 .000      

ADV 2.521 .182 13.868 .000 .891 .878 .731


BONUS 1.981 .478 .554 .481 .219
a. Dependent Variable: SALES

Question 2.1
What are the values of zero-order, partial and part correlation for ADV in Model 1 above?
Explain.

Question 2.2
What percentage of the variation in SALES is explained by Model 2?
18

Question 2.3
Conduct an appropriate hypothesis test at a 95% confidence level to decide if BONUS is a
significant variable? State the hypothesis clearly.

Question 2.4
Which of the variables, ADV or BONUS, has a greater impact on SALES in Model 2? Show all
work.

Question 2.5
Compare the coefficient of ADV in Models 1 and 2 and comment on the decrease in the value.
What is the likely reason for the decrease?

Question 2.6
Calculate the VIF for ADV and BONUS in Model 2. What do these values imply?

Meddicorp’s sales manager believes that in addition to ADV and BONUS, two other variables
might be important in explaining the variation in SALES. These are:

MKTSHR – Market share currently held by Meddicorp in each territory


COMPET – largest competitor’s sales in each territory

A regression model is developed (Model 3) using the ENTER method to predict SALES with all the
4 explanatory variables. The SPSS model summary is given below:

Model 3 Summaryb

Model R R Square Adjusted R Std. Error of the


Square Estimate

.919a .844 .833 90.86720

a. Predictors: (Constant), COMPET, BONUS, MKTSHR, ADV


b. Dependent Variable: SALES
Model 3 ANOVAa

Model Sum of Squares df Mean Square F Sig.

Regression 2460583.610 4 615145.903 74.501 .000b

Residual 454126.635 55 8256.848

Total 2914710.246 59

a. Dependent Variable: SALES


b. Predictors: (Constant), COMPET, BONUS, MKTSHR, ADV

Model 3 Coefficientsa
19

Standardize
d
Unstandardized Coefficients Coefficients
Std.
Model B Error Beta t Sig.
(Constant) -662.385 170.294   -3.890 .000
BONUS 1.997 .483 .240 4.137 .000
ADV 2.523 .204 .797 12.395 .000
MKTSHR 2.761 3.058 .051 .903 .370
COMPET -.009 .237 -.003 -.040 .968
a. Dependent Variable: SALES

Question 2.7
Perform an appropriate hypothesis test at 95% confidence level to determine if in addition to
ADV and BONUS, the variables MKTSHR and COMPET are important in explaining the variation in
SALES.

A 4th model is developed (Model 4) taking COMPET, BONUS and ADV as the independent
variables, and the regression summary is given below:

Model 4 Summaryb

Model R R Square Adjusted R Std. Error of the


Square Estimate

.918a .842 90.71747

a. Predictors: (Constant), COMPET, BONUS, ADV


b. Dependent Variable: SALES
Model 4 Coefficientsa

Standardize
d
Unstandardized Coefficients Coefficients
Std.
Model B Error Beta t Sig.
(Constant) -566.744 133.140   -4.257 .000
BONUS 1.979 .481 .238 4.110 .000
ADV 2.552 .201 .806 12.715 .000
COMPET -.083 .222 -.022 -.372 .711

a. Dependent Variable: SALES

Question 2.8
Calculate the Adjusted R Square value for Model 3. Compare the R square and Adjusted R
square values for Models 2 and 3 and comment on the change in these values.
Question 2.9
Which of the 4 models above is the most appropriate for predicting SALES? Explain clearly.
20

Question 2.10
Use the model chosen above to predict SALES given:
ADV = 530, BONUS = 300, Stamp = 5, MKTSHR = 25, COMPET = 350

Question 3

In a medical research study, 36 patients suffering from severe clinical depression were
prescribed different treatments and the effectiveness of these treatments in managing severe
clinical depression was recorded. Note that for the duration of the study, each patient was
uniquely prescribed one and only one of the four available treatments (referred to as
Treatments A, B, C, and D). Moreover, each treatment was prescribed to the exact same number
of patients. Finally, all of the variables recorded in the study are given in Table 3.1.

Table 3.1. Variable names, descriptions, and types

Variable Variable Description Variable Type


The effectiveness of the prescribed
Effectiveness Numerical
treatment measured on a scale of 0-100
Age Age of the patient Numerical
Gender Gender of the patient Categorical (Male/Female)
Treatment Type of treatment prescribed to a patient Categorical (A, B, C, D)

The pairwise correlation matrix for all these variables is first computed as shown below (Table
3.2):

Table 3.2 Correlation matrix

Treatment Treatment Treatment Treatment


Effectiveness Age Gender
A B C D
Effectiveness 1
Age 0.7967 1
Gender 0.2262 0.0721 1
Treatment A 0.3589 0.0355 0.1360 1
Treatment B -0.1231 0.0667 - 0.1360 - 0.3333 1
Treatment C -0.2489 -0.0756 0.1360 - 0.3333 - 0.3333 1
Treatment D 0.0131 -0.0266 - 0.1360 - 0.3333 - 0.3333 - 0.3333 1

Question 3.1
21

As the correlation coefficient value between Effectiveness and Treatment C is negative, it was
concluded that Treatment C is not effective in treating clinical depression. State whether this
claim is true or false and justify your conclusion.

Model 1:

A linear regression model was initially constructed to predict effectiveness based only on the
treatment used and the following SPSS output was obtained.

Model 1 Coefficientsa

Model Unstandardized Coefficients Standardized t Sig.


Coefficients

B Std. Error Beta

(Constant) 55.444

TreatmentA 7.333 .259


1
TreatmentB -2.889 -.102

TreatmentC - 5.555 -.197

a. Dependent Variable: Effectiveness

Question 3.2.
Using the regression coefficients given in Model 1, rank the different treatments in the order of
their effectiveness, starting from best to worst and clearly indicate the average effectiveness of
each treatment.

Question 3.3.

What would the prediction equation be if the base category was taken to be Treatment B?

Question 3.4.
It was noted that the total variability in effectiveness explained by the above regression model is
only 15.46%. Moreover, when a reputed clinical psychologist was shown the rankings of the
treatments, she stated that based on her clinical experience all treatments are likely to have the
same effectiveness on average. Perform an appropriate test to verify this claim at a 95%
confidence level, clearly stating the null and alternate hypotheses.

Model 2
22

Recognizing that ‘Age’ of the patient might be a factor when measuring the effectiveness of a
received treatment, Model 1 was enhanced by using an interaction variable, namely
“Age*Treatment” (all combinations) and the regression output for this new model is given
below.
Model Summaryb

Model R R Square Adjusted R Std. Error of the


Square Estimate

1 .937a .878 .848 4.84794

a. Predictors: (Constant), Age_TreatD, Age_TreatC, Age_TreatB,


Age_TreatA, TreatmentC, TreatmentD, TreatmentB,
b. Dependent Variable: Effectiveness
Coefficientsa
Model Unstandardized Coefficients Standardized t Sig.
Coefficients
B Std. Error Beta
(Constant) 49.470 5.864 8.437 .000
TreatmentB -22.030 7.882 -.779 -2.795 .009
TreatmentC -44.931 7.512 -1.589 -5.981 .000
TreatmentD -20.964 7.844 -.742 -2.673 .012
1
Age_TreatA .296 .125 .496 2.361 .025
Age_TreatB .549 .110 .948 5.010 .000
Age_TreatC 1.074 .104 1.742 10.286 .000
Age_TreatD .620 .114 1.018 5.439 .000
a. Dependent Variable: Effectiveness

Question 3.5.
Based on the regression output, plot the regression equations for each of the four treatments
(on the same graph), if the minimum and maximum ages of the patients in the dataset are 19
and 67 years, respectively. Based on your regression plots, which treatment would you
recommend as the best on average for a patient of age 60 years?

Question 3.6.
At what age does the average effectiveness of Treatment C surpass that of Treatment A? Based
on your answer, what treatment(s) would you recommend for younger patients and older
patients?

Question 3.7.

What is the probability that the effectiveness level would be at least 60 for a 55-year old patient
who is administered Treatment B?

Model 3
23

A stepwise regression procedure was initiated to predict effectiveness using all the original
explanatory variables given in Table 3.1 as well as interaction variables: Age*Treatment A;
Age*Treatment B; and Age*Treatment C. The following SPSS output obtained in the first two
models of the stepwise regression procedure is shown below.
Variables Entered/Removeda
Step Variables Entered Variables Removed Method
Stepwise (Criteria: Probability-
of-F-to-enter <= .050,
1 Age . Probability-of-F-to-remove >= .
100).

Stepwise (Criteria: Probability-


of-F-to-enter <= .050,
2 TreatmentA . Probability-of-F-to-remove >= .
100).

a. Dependent Variable: Effectiveness

Question 3.8

At the end of Step 1 of the regression procedure, determine the partial and part correlations for
Treatment A. Correspondingly, determine what percentage of the variability in “effectiveness” is
explained by Step 1 and Step 2, respectively?

Question 3.9
After Step 2 of the regression model, the excluded variable with the largest semi-partial (part)
correlation value = 0.219, was “Age*Treatment A”. Conduct an appropriate test at the 95%
confidence level to determine if this excluded variable can be added to the regression model.
State the null and alternate hypotheses and show all calculations.

Question 3.10
In Model 3, only three interaction variables, namely Age*Treatment A; Age*Treatment B; and
Age*Treatment C are considered as possible explanatory variables. Why is the fourth interaction
variable, namely “Age*Treatment D” not included in this case, while it was included in Model 2?
Explain.

Question 1
24

Per Capita Income of 20 countries were analysed using the variables described in Table
1.

Table 1. Data Dictionary


S.No Variable Variable Type Code in SPSS output
1 Per Capita Income (in Numerical Per Capita
Dollars)
2 Corruption Index Integer CI
(Higher Value indicates
lower level of corruption
in the country)
3 Gini Index (Measure of Numerical Gini
Wealth Distribution and
Discrimination)
4 Communist State Binary CS
(Whether the county 1 = Communist State; 0
was/is a communist otherwise
state)

Descriptive statistics of the variables and correlations are shown in Tables 2 and 3
respectively.

Table 2 Descriptive Statistics


Std.
N Minimum Maximum Mean
Deviation
CI 20 29.0 90.0 61.700 20.6171
Gini 20 23.5 53.7 34.740 7.3846
CS 20 .0 1.0 .250 .4443
Per Capita 20 12275.0 69249.0 37789.050 15847.4829
Valid N
20
(listwise)

Table 3 Correlations
CI Gini CS PerCapita
CI 1 -.464* -.612** .862**
Gini -.464* 1 .253 -.338
CS -.612** .253 1 -.556*
Per Capita .862** -.338 -.556* 1

Model 1
Y (Per Capita) = 0 + 1 x CI
25

A simple linear regression model (Model 1) is developed between per capita (Y) and
corruption index (CI). SPSS model outputs are shown in Tables 4 and 5. Normal P-P
Plot and Residual Plot are shown in Figures 1 and 2 respectively.

Table 4 Model Summaryb


Model R R Square Adjusted R Std. Error of the
Square Estimate
1 8241.4390
a. Predictors: (Constant), CI
b. Dependent Variable: Per Capita

Table 5 Coefficientsa
Model Unstandardized Coefficients Standardized t Sig.
Coefficients
B Std. Error Beta
(Constant) -3112.753 5950.818 -.523 .607
1
CI 662.914 91.706 .862
a. Dependent Variable: Per Capita

Figure 1 Normal Probability Plot Figure 2 Residual Plot

Question 1.1
What proportion of the variation in per capita income is explained by corruption index
(CI)?

Question 1.2
26

Is there a statistically significant relationship between corruption index and per capita
income of the countries at 5% significance?

Question 1.3
Is it possible to conclude that the per capita income increases by at least 500 dollars for
every one unit increase in corruption index at 10% significance level? Clearly write all
the steps.

Question 1.4
What can you conclude about model between per capita and CI based on the plots in
Figures 1 and 2?

Question 1.5
What is the minimum average value of per capita at 95% confidence interval when CI =
50?

A second model is developed between Per Capita and Communists States (CS).
Model 2
Y (Per Capita) = 0 + 1 x CS
The output is shown in Table 6. Normal probability and residual plots are shown in
Figures 3 and 4 respectively.
Table 6 Coefficientsa
Model Unstandardized Coefficients Standardized t Sig.
Coefficients
B Std. Error Beta
(Constant) 42743.933 3495.319 12.229 .000
1
CS 6990.639 -.556
a. Dependent Variable: PerCapita
27

Figure 3 Normal Probability Plot


Figure 4 Residual Plot

Question 1.6
Calculate the average per capita of communist states. Clearly write all the steps?

Question 1.7
Is model 2 statistically significant at 5% significance, use all the information (Table 6,
Figures 3 and 4) provided. Clearly write all the arguments.

A stepwise regression model is developed using CI and Gini as independent variables and
the outputs are shown in Table 7.
Table 7 Stepwise Regression Output
Model Unstandardized t Sig. Correlations
Coefficients
B Std. Error Zero-order Partial Part
(Constant) -3112.753 5950.818 -.523 .607
1
CI 662.914 91.706 7.229 .000 .862 .862 .862
(Constant) -10781.284 14572.250 -.740 .469
2 CI 691.235 105.487 6.553 .000 .862 .846 .797
Gini -.338 .139 .070

Question 1.8
What is the value of R-square after adding the variable Gini to the model?

Question 1.9
28

Carry out an appropriate hypothesis test to check whether the variable “Gini” is worth
adding to the model at 5% significance.

Question 1.10
Calculate the variance inflation factor between variables CI and Gini. What can you
conclude from the calculated VIF value?

A stepwise regression model is developed using all the 3 independent variables and the
SPSS outputs are given in Tables 8 and 9
Table 8 Coefficientsa
Model Unstandardized Coefficients Standardized t Sig.
Coefficients
B Std. Error Beta
(Constant) -3112.753 5950.818 -.523 .607
1
CI 662.914 91.706 .862 7.229 .000
a. Dependent Variable: PerCapita

Table 9 Excluded Variablesa


Model Beta In t Sig. Partial Collinearity
Correlation Statistics
Tolerance
Gini .079b .579 .570 .139 .785
1 b
CS -.044 -.287 .777 -.070 .625
a. Dependent Variable: PerCapita
b. Predictors in the Model: (Constant), CI

Question 1.11
Based on the information provided in Tables 8 and 9, is it possible to conclude that there
is no statistically significant relationship between Per Capita and independent variables
Gini and CS? Excluded variables in Table 9 are variables that are not part of the
regression model (statistically not significant) when stepwise regression is used.

Question 2
Applicants who apply for a job at Precision Watches Inc., which requires extensive
manual assembly of small intricate parts, are initially given three different tests to
measure their manual dexterity. The ones who are hired are then periodically given a
performance rating on a 0-100 scale that combines their speed and accuracy in
performing the required assembly operations. Data is collected on the test scores and
performance ratings for a randomly selected group of 80 employees who continued
working for the company. Their seniority (months with the company) at the time of the
29

performance rating is also noted. The summary information and the results from four
regression models developed using the data are given below:

Pairwise Correlation Matrix


  JobPerf Seniority Test1 Test2 Test3
JobPerf 1
Seniority 0.43 1.00
Test1 0.58 1.00
Test2 0.52 0.60 1.00
Test3 0.62 0.66 0.80 1.00
Descriptive Statistics
Minimu Maximu Std.
N m m Mean Deviation
JobPerf 80 38 100 65.75 10.630
Seniority 80 7 30 18.89 5.00
Test1 80 31 82 60.53 9.576
Test2 80 37 86 60.75 9.872
Test3 80 26 77 50.71 9.181
Valid N
80
(listwise)

Model 1 Summary
Std. Error
R Adjusted of the Durbin-
Model R Square R Square Estimate Watson
1 .176 9.651 1.856
a Predictors: (Constant), Seniority
b Dependent Variable: JobPerf

Model 1 ANOVA
Sum of Mean
Model Squares Df Square F Sig.
1 Regression 1662.584 1 1662.584 17.852 .000
Residual 7264.416 78 93.134
Total 8927.000 79

Model 1 Coefficients

Unstandardized Standardized
Model Coefficients Coefficients t Sig.
B Std. Error Beta
1 (Constant) 11.8
48.928 4.125 .000
61
Seniority .891 .432
30

Model 2 Summary
Adjusted Std. Error
R R of the Durbin-
Model R Square Square Estimate Watson
1 .764 .583 .561 7.042 1.878
a Predictors: (Constant), Test3, Seniority, Test1, Test2
b Dependent Variable: JobPerf

Model 2 ANOVA
Sum of Mean
Model Squares Df Square F Sig.
1 Regression 5208.110 4 1302.027 26.258 .000
Residual 3718.890 75 49.585
Total 8927.000 79

Model 2 Coefficients
Standardize
Mode Unstandardized d Collinearity
l Coefficients Coefficients t Sig. Statistics
Std. Toleranc
B Error Beta e VIF
1 (Constant) 6.557 6.187 1.060 .293
Seniority .801 .155 .388 5.171 .000 .986 1.014
Test1 .300 .112 .271 2.693 .009 .550 1.819
Test2 .086 .135 .080 .640 .524 .355 2.816
Test3 .407 .154 .352 2.638 .010 .313 3.197

Model 3 Summary
Adjusted Std. Error
R R of the Durbin-
Model R Square Square Estimate Watson
1 .762 .581 .565 7.014 1.891
a Predictors: (Constant), Test3, Seniority, Test1
b Dependent Variable: JobPerf

Model 3 ANOVA
Sum of Mean
Model Squares Df Square F Sig.
1 Regression 5187.803 3 1729.268 35.148 .000
Residual 3739.197 76 49.200
Total 8927.000 79

Model 3 Coefficients
31

Collinearity
Standardize Statistics
Unstandardized d
Model Coefficients Coefficients t Sig.
Std. Toleranc
B Error Beta e VIF
1 (Constant) 7.893 5.801 1.361 .178
Seniority .793 .154 5.157 .000 .993 1.008
Test1 .312 .110 2.844 .006 .565 1.771
Test3 .473 .114 4.145 .000 .567 1.764

Model 4 Summary
Adjusted Std. Error
R R of the Durbin-
Model R Square Square Estimate Watson
1 .757 .574 .562 7.031 1.843
a Predictors: (Constant), AvgScore, Seniority
b Dependent Variable: JobPerf

Model 4 ANOVA
Sum of Mean
Model Squares Df Square F Sig.
1 Regression 5120.011 2 2560.006 51.779 .000
Residual 3806.989 77 49.441
Total 8927.000 79

Model 4 Coefficients
Standardize
Mo Unstandardized d Collinearity
del Coefficients Coefficients t Sig. Statistics
Std. Toleranc
B Error Beta e VIF
1 (Constant) 5.407 6.010 .900 .371
Seniority .821 .154 .398 5.339 .000 .997 1.003
AvgScore .782 .094 .623 8.362 .000 .997 1.003

Use the information given above to answer the following questions. Specify the model(s)
you use to draw your conclusions, where relevant.

a) Can it be concluded that performance rating improves with length of stay with the
company (Seniority), irrespective of the original test scores? Select the appropriate
model to answer the question.
32

b) Predict the average performance rating for a worker who has 15 months of Seniority.
What are the highest and lowest performance ratings that this worker is likely to get
at 90% confidence level?

c) If Test 2 was used to predict performance scores on its own, is it likely to be a


significant predictor of JobPerf? Justify. In the presence of other 3 variables is it a
significant predictor. Why or not why not?

d) Can it be concluded that employees with higher average scores on the tests stay
longer with the company? Choose the appropriate models to compare.

e) Two employees whose seniority differs by 5 months have the same average test score.
Can it be concluded that the performance rating of the more senior employee will be
at least 3 points higher at 5% significance level?

Question 3
A data analytics start up works with political parties during elections. They have got
access to voting patterns from various official sources. They are trying to understand how
the percent of votes obtained by the winner is determined. As a first cut they are using the
following data:

% VOTES – the percent of votes polled obtained by the winning candidate


MARGIN – the margin of victory measured in number of votes
Gender – 1 is for Men and 0 for women
College – 1 is for college educated winners and 0 for those who did not go to college.
They run the regression for all 543 elected MPs. The model output is provided below
(with few missing information):

Table 3.1
Regression
Statistics  
Multiple R
R Square
Adjusted R
Square
Standard
Error
Observations 543
33

Table 3.2
ANOVA
Significanc
  df SS MS F eF
Regression
Residual 17104.06    
Total 542 36481.89      

Table 3.3 Coefficients


Coefficient Standard P-
  s Error t Stat value Lower 95% Upper 95%
Intercept 38.59235 0.937225 36.75129 40.4334106
MARGIN 5.32E-05 2.18E-06 4.89E-05 5.7463E-05
Gender 1.551306 0.777806 0.023404 3.07920835
College -1.47506 0.586995 -2.62814 -0.3219783

(i) Fill up the Tables 3.1, 3.2 and 3.3 above (except the p values and the Significance
F values). Clearly write all the steps.

(ii) Assuming that t is significant for any value greater than 1.964 at 5%, are the
variables (margin, gender and college) significant?

(iii) Assuming that the critical value of F is 2.621 at 5% significance, is the overall
regression significant?

The analytics firm decides to dig a little deeper and looks at two outlying states, UP and
AP, one of which has significantly lower assets per winner and the other significantly
higher. Both the new variables are 0-1 variables. The values for some of the regressions
are given below (Table 3.4).

Table 3.4 Regression Models with Corresponding R-Square


Regression
Model Independent Variables R2
1 MARGIN
2 MARGIN, Gender 0.52567
3 MARGIN, Gender, College 0.531163
MARGIN, Gender, College,
4 UP 0.56051
MARGIN, Gender, College,
5 UP, AP 0.581339
34

(iv) What is the part correlation for College and % of votes in Regression model 3?

(v) Between regression 2 and 5 is it justified to add the additional variables?

Regression model 5 in Table 3.4 has a standard error of 5.333135, an overall F value of
149.1324 with significance of 4.4x10-99 . The standard deviation for the dependent
variable is 8.204253. The values of standard deviation for the dependent and independent
variables are given below (Table 3.5).

Table 3.5
Coefficient Standard
  s deviation
Intercept 38.56993  
MARGIN 5.58E-05 111365.7
Gender 1.498308 0.311494
College -1.53774 0.412796
UP -3.71439 0.354761
AP 5.715821 0.209766

(vi) Which variable has the greatest impact on Voting % ?


35

End-term:

Question 1 (10 marks)

Sholay Gaadiwale, Inc., a large automobile manufacturing company, manufactures three


different families of vehicles: family of trucks; family of compact sedans; and family of
midsize/luxury coupes and sedans. One assembly plant outside Ramgarh, KA, assembles
two models from the family of midsize/luxury coupes and sedans.

The “Jay” model is a two-door luxury coupe with leather seats, wooden interiors,
customized features, and navigational capabilities. It is marketed for high-net-worth
individuals and generates a healthy $5400/car profit for the company. The second model,
named “Veeru”, is a four-door entry-level sedan with vinyl seats, plastic interiors, and
standard features. It is targeted towards mid-career professionals with families and
generates a modest $3600/car profit for the company.

Basanti Ghodewali, the Vice President of Sholay, Inc., is currently deciding the
production schedule for the next month. The various components required for assembly
of the two models are shipped from various manufacturing plants across the country. She
estimates that the assembly plant has a capacity of 48000 labor-hours for the duration of
the month, and it is also known that one Jay car requires 12 labor-hours of assembly time
and one Veeru car requires 6 labor-hours of assembly time.

Thakur, the forecasting expert, has shared with Basanti that they will only be able to
procure 20000 doors (10000 right-hand side doors and 10000 left-hand side doors) from
their car door manufacturing plant. Both Jay and Veeru cars use the same doors.
Moreover, Thakur has also forecasted that the demand for the Jay model will be at most
3000 cars for next month, whereas the demand for the Veeru model is only limited by the
number of cars assembled.

Basanti wishes to run through some scenarios before making any decisions. Towards this
end, answer the following questions:

(a) Formulate a linear programming problem to determine the optimal number of Jay and
Veeru models that need to be assembled at the plant in order to maximize profit.

(b) The graphical representation of the linear program formulated in part (a) is given
below. Identify the coordinates of all the extreme points (labeled 1-5) and determine
the optimal product mix and the optimum objective function value.
36

(c) Ram Sippy, the Chairman of the Board of Directors for Sholay, Inc., wishes to
capture a larger share of the luxury car market, and therefore wishes to meet the full
demand for the Jay model. By how much would the profit of the assembly plant
reduce as compared to part (b) if they pursue this strategy? Ram Sippy has indicated
that the company will take on this strategy if the profit reduction is lesser than
$1,500,000. Should the company adopt this strategy?

(d) Gabbar, the assembly plant foreman knows that as per the latest round of negotiations
with the labor union, management can demand an increase in the labor-hour capacity
as and when required. How much additional profit can be made as compared to part
(b) for a unit increase in labor-hours? What is the maximum increase in labor-hours
for which this increase in profit is valid?
37

Question 2 (15 marks)

The Wonder Sports Supply Company produces All-Pro (A) and College (C) footballs at
its manufacturing plant in Jalandhar. The maximum demand from wholesalers for the
months of February and March is as follows:

Jalandhar
Product
February March
A 3600 6300
C 4500 5400

The plant has 20 and 23 production days available in February and March, respectively,
and an inventory capacity of 1000 footballs (for A and C models combined) is available
in any given month. The inventory holding cost is Rs. 30/- per unit for A and Rs. 40/- per
unit for C.
There are two production processes that can be used at the plant to produce the footballs
at the following production costs (Rs. per unit):

Plant 1
Product
Process 1 Process 2
A 620 590
C 780 850

The number of footballs that can be produced per day varies according to the process and
are as given below:
Plant 1
Product
Process 1 Process 2
A 100 140
C 120 150

The net sales revenue (selling price – shipping costs) are Rs. 830/- per unit for A and Rs.
1120/- per unit for C. Wonder Sports would like to determine the optimal production and
sales plan that will maximize its total profit.
(a) Does Wonder Sports have enough capacity to meet all the demand?
(b) Define the decision variables required to determine the optimal production and sales
plan for Wonder Sports.

(c) State the constraints and label them clearly.


(d) State the objective function.
38

Question 3 (20 marks)

Ranchoddas Shamaldas Chanchad (Rancho) established CNG (Cleanliness is Next to


Godliness), a manufacturing firm that specialized in manufacturing hand sanitizers in
2020. Sanitizers manufactured by CNG are made of three main ingredients: (i)
Benzalkonium Chloride (BC); (ii) Ethyl alcohol (EA); and (iii) Isopropyl alcohol (IS).
CNG makes three different varieties of hand sanitizers namely: Regular, Premium, and
OCD. The ingredients used (in ml) to produce one litre of sanitizer is shown in the
following table, along with demand for different sanitizers and availability of ingredients.

Ingredient in ml for 1 litre of Sanitizer Minimum


Profit from
Demand for
Sanitizer Sanitizer
BC EA IS Sanitizer (in
(Rs/litre)
litres)
Regular A11 600 200 75 40
Premium 100 300 450 80 60
OCD 100 400 A33 20 100
Availability of
ingredients (in 20000 80000 100000
ml)

The corresponding LP which maximizes profit subject to resource availability and


demand constraints is formulated below:

Maximize

subject to:

(constraint for
BC)

(constraint for
EA)

(constraint for
IS)

(constraint for Regular


Sanitizer)
39

(constraint for Premium


Sanitizer)

(constraint for OCD


Sanitizer)

The Excel Solver sensitivity output is given below:

Variable Cells
Final Reduced Objective Allowable Allowable
Cell Name Value Cost Coefficient Increase Decrease
$A$
X1 75 0 40 110 1E+30
2
$B$
X2 80 0 60 15 1E+30
2
$C$
X3 27.5 0 100 1E+30 20
2

Constraints
Final Shadow Constraint Allowable Allowable
Cell Name Value Price R.H. Side Increase Decrease
$A$
BC 14500 20000 1E+30 5500
4
$A$
EA 80000 0.25 80000 22000 3000
5
$A$
IS 64750 100000 1E+30 35250
6
$A$
Regular 75 75 5 55
7
$A$
Premium 80 80 10 80
8
$A$
OCD 27.5 20 7.5 1E+30
9

(a) Determine the values of A11 and A33 in the LP formulation.

(b) What is the shadow price for the Benzalkonium Chloride constraint?
40

(c) Write the dual to the LP primal problem formulation provided above. Using the
complementary slackness theorem, calculate the shadow price values of
constraints, Regular and Premium.

(d) Due to Covid-19, the profit earned from regular sanitizer has increased from 40 to
100, what will be the impact of this change on the optimal solution and the profit?

(e) CNG was informed that one of the suppliers of Ethyl Alcohol can supply
additional 20000 ml of EA. Should CNG buy this additional quantity of EA?
What will be the impact of this additional quantity on the profit?

(f) Due to increased demand for sanitizers, CNG has increased the price such that the
profit of regular is 60, premium is 90 and OCD is 150. What will be the impact of
this increase on the profit?

(g) There is a demand for 5000ml BC in the open market at INR 2 per ml. Should
CNG sell BC directly in the market?

(h) CNG would like to introduce a new Luxury Sanitizer, which requires 200ml of
BC, 500ml of EA and 400ml of IS. The profit from this new Sanitizer is 140.
Should CNG introduce the new Sanitizer?

Question 4 (15 marks)

The Board of Directors for Cobra, Inc. is currently evaluating seven large potential
business ventures. The amount of capital investment required to participate in these
business ventures as well as the profit they yield after three years (in millions of dollars)
is specified in Table 1 below.

Table 1. Required capital and profit yield for seven business ventures
Business Venture
1 2 3 4 5 6 7
Capital Required 43 28 34 48 17 32 23
Estimated Profit 17 10 15 19 7 13 9

Cobra Inc. has a total amount of $100 million of capital available for investment into
these new business ventures. Using this information, answer the following questions:
41

(a) Formulate an integer program to determine which of these business ventures should
be selected by Cobra Inc. to maximize their total profit at the end of three years
subject to their budget constraint. Clearly define your decision variables, and then
write the constraints and the objective function.

(b) Using MS Excel Solver, determine the optimal solution to the problem. Upload your
MS Excel file showing the formulation and calculations.

(c) If Cobra Inc. invests in Venture 3, then they must invest in Venture 5. What
constraint needs to be added to the formulation in part (a) to reflect this condition?
Will this constraint change your optimal solution from part (b)?

(d) Suppose that Cobra Inc. can invest either in Venture 3 or Venture 4 but not both.
What constraint needs to be added to the formulation in part (a) to reflect this
condition? Will this condition change the optimal solution from part (b)?

(e) If Cobra Inc. invests in at least two of Ventures 1, 2, 3, or 4 then they must invest in
at least one of Ventures 5, 6, or 7. What constraint needs to be added to the
formulation in part (a) to reflect this condition?

(f) If Cobra Inc. invests in at least two of Ventures 1, 2, or 3, then they must invest in at
most two of Ventures 4, 5, 6, or 7. What constraint needs to be added to the
formulation in part (a) to reflect this condition?
42

Question 5 (15 marks)

Consider the following integer programming problem:

Minimize
subject to:

You are given a partially completed branch-and-bound tree below, where the labels at the
center of the nodes indicate the order in which the nodes were evaluated, i.e., its LP
relaxation was solved. The notation indicates that a linear programming
relaxation at the node labeled k was solved and the objective function value is q. The
notation indicates that the solution to the LP relaxation at node k is
given by , , , . An active node is defined as one that still
needs to be explored and cannot be fathomed. At any stage, the incumbent solution is the
best-known integer (feasible) solution to the problem.

You are required to complete the branch-and-bound tree and determine the optimal
solution to the problem. Towards this end, answer the following questions:

(a) What is the current incumbent solution?

(b) What is the best lower bound on the problem at this stage?

(c) What are the active nodes in the branch-and-bound tree at this stage?

(d) Formulate and use MS Excel Solver to solve the linear programming relaxation at
Node 5. Upload your MS Excel file showing the formulation and calculations.
43

(e) Based on your solution to part (d), did the incumbent value change? If yes, what is
the new incumbent solution?

(f) What are the active nodes in the branch-and-bound tree at this stage?
(g) From the solution shown at Node 6, select an appropriate branching variable and
determine the optimal solution at Node 8. Did your incumbent solution change?

(h) Can the branch-and-bound process be terminated at this stage? If yes, explain your
reasoning. If not, continue branching and determine the optimal solution to the
problem.
44
45

Question 1

Rangaswamy owns 50 acres of land in the Hassan district of Karnataka on which he


grows bananas and papayas. The total cost of producing bananas is Rs. 1,05,000 per acre
while the total cost for papayas is Rs. 2,10,000 per acre. For the coming year
Rangaswamy has a budget of Rs. 63,00,000. He does not want to use any more than 15
acres for bananas and plans on using at least 20 acres for papayas. The profit from each
acre of bananas is Rs. 80,000 and from papayas is Rs. 1,40,000.

a) Formulate a linear programming problem (call this Primal LPP) to help


Rangaswamy decide how many acres of each fruit to produce in order to
maximize his total profit. Clearly define all the decision variables, objective
function and constraints.

b) Use the graphical method to find the optimal solution and optimal objective
function value. Use the graph paper provided and show all work on the graph.

Optimal Solution:

Optimal Objective Function Value:

c) How many acres of land should Rangaswamy plan on cultivating in the coming
year?

d) It is expected that the profit per acre from bananas will drop in the coming year
due to over-production of bananas in other regions. To what level can the profit
per acre from bananas drop without affecting Ramaswamy’s optimal production
plan?
How many acres of bananas should be produced if the profit drops below this
level?

e) Due to a shortage of papayas, the profit will increase to Rs. 1,60,000 per acre.
What are the various production plans that will maximize Ramaswamy’s total
profit?
46

f) Does Rangaswamy stand to make a higher profit if he relaxes the maximum limit
of 15 acres he plans on using for producing bananas? Why or why not? If yes,
what would he gain per additional acre that he allows for producing bananas?
How many additional acres will this additional gain per acre be valid for?

Question 2
David Dyewala (DD) uses 4 types of pigments, namely, Aluminium, Copper, Manganese
and Gold to dye various fabrics sold by them in the market. Using these pigments DD
produces 4 colours of fabric, namely, Sarcoline, Mikado, Glacous and Xanadu. The
numbers in the Table below indicate the proportions of the 4 pigments required for
dyeing the fabrics in the given colours:

Table 2. Proportion of pigment required for the colours produced


Pigments
Colours
Aluminium Copper Manganese Gold
Sarcoline 2 2 3 1
Mikado 1 2 2 1
Glacous 2 2 3 2
Xanadu 1 1 1 2

For dyeing one square meter of the fabric the total quantity (in grams) of pigments
required is as given below:

Table 3. Total Quantity of pigments required to dye one square meter of fabric
Total quantity of pigments
Colours
required (in grams)
Sarcoline 200
Mikado 120
Glacous 180
Xanadu 150

The supply of pigments is limited to 50 kgs for Aluminium, 40 kgs for Copper, 80 kgs for
Manganese and 20 kgs for Gold. The profit earned from one square meter of fabric
coloured Sarcoline, Mikado, Glacous and Xanadu is 80, 120, 60 and 140, respectively.
The quantity of Xanadu coloured fabric produced should be twice that of the remaining
colours together.

1. Write a linear programming model to maximize the profit. Write clearly all the
decision variables and constraints.
47

2. It is expected that during the dyeing process there could be 10% wastage of
pigments. Modify the formulation suitably to meet this additional constraint.

3. Due to environmental regulations, there is additional cleaning cost of rupees 20


per square meter if the fabric produced of any colour exceeds 10 square meters.
Make appropriate changes to the formulation to incorporate this additional cost.

Question 3
George Kutty (GK) is the co-founder and CEO of George Kutty Cookies (GKC), a
cookie manufacturing company based out of Palakaad, Kerala. GKC specializes in
producing four different varieties of cookies namely: (1) Butter Cookies (BC); (2) Low
Calorie Cookies (LC); (3) Fiber Cookies (FC); and (4) Short Bread Cookies (SB).
The main ingredients required in the production of these cookies are: (1) Dough (made of
wheat flour); (2) Butter; (3) Sugar; and (4) Preservatives. The quantity of various
ingredients (per kilogram of cookie) for various cookies is given in Table 3 below.

Table 4. Quantity of ingredients required for one kilogram of cookies


Quantity in Kg per Kg of cookies manufactured
Cookies Dough Butter Sugar Preservative
Butter Cookies (BC) x1 0.80 0.5 0.75 0.10
Low Calorie (LC) x2 1.20 0.2 0.25 0.30
Fiber cookies (FC) x3 0.75 0.6 0.50 0.2
Short Bread (SB) x4 1.25 0.80 0.80 0.15

The profit per kilogram earned from cookies BC, LC, FC and SB are 120, 100, 90 and
150, respectively. The maximum demand (in kgs) for BC, LC, FC and SB are 200, 150,
120 and 80, respectively. Due to floods in Kerala during August 2018, GKC is unable to
receive supply of ingredients and has to manage the production with the existing
inventory of ingredients. The inventory (in kgs) available for Dough, Butter, Sugar, and
Preservatives are 250, 180, 200 and 40, respectively. The following LP formulation is
written to maximize the profit for GKC, where the decision variables used are as defined
below.
= Amount (in kgs) of BC produced
= Amount (in kgs) of LC produced
= Amount (in kgs) of FC produced
= Amount (in kgs) of SB produced

Maximize:
subject to:
48

The Solver output for the above formulation is provided below:


Final Reduced Objective Allowable Allowable
Cell Name
Value Cost Coefficient Increase Decrease
3.78378378
$A$2 x1 200 120
4
8.23529411
$B$2 x2 14.6718 100 44
8
$C$2 x3 0 90
2.66666666
$D$2 x4 57.9150 150 3.5
7

Final Shadow Constraint Allowable Allowable


Cell Name
Value Price R.H. Side Increase Decrease
$A$4 Dough 250 250 22.0987 11.875
$A$5 Butter 149.2664 180 1E+30 30.7335
$A$6 Sugar 200 84.9420 200 7.6 22.9487
$A$7 Preservative 33.0888 40 1E+30 6.9111
$A$8 BC 200 200 53.5714 20.4285
$A$9 LC 14.6718 150 1E+30 135.3281
$A$10 FC 0 120 1E+30 120
$A$11 SB 57.9150 80 1E+30 22.0849

1. Calculate the maximum profit that GKC can earn.

2. GKC can purchase additional preservative from the local market at rupees 20 per
kg. Should GKC purchase additional preservatives?

3. Another cookie manufacturer in the city has agreed to provide 20 Kgs of dough
for rupees 50 per kg. Should GKC buy this additional dough?

4. Travancore airlines has requested GKC to supply Fiber cookies. Should GKC
supply fiber cookies to the airline at the current profit? What will be the impact
49

on the profitability if GKC decides to supply fiber cookies to the airline?

5. GKC decides to increase the profit on low calorie cookies to 140. What will be
the impact of this change on the optimal solution?

6. There is a market demand for 30Kg of butter for INR 80 per Kg. Should GKC
take this offer and sell 30 Kg of butter directly in the market instead of using in
the production of cookies?

7. Production manager feels that there could be a production wastage of up to 5 Kg


of dough and 10 kg of sugar. What will be the impact of these wastages on the
profit?

8. GKC can produce a new variety called “sugar free cookies” using 1.5 Kg of
dough, 0.5 kg of butter and 0.3 kg of preservatives. If the profit that can be
earned from sugar free cookies is rupees 180 per kg, comment whether GKC
should produce sugar free cookies?

Question 4

Consider a General Assignment Problem (GAP) problem, where a set of


jobs are to be assigned to a set of machines.
The 6 jobs require 7, 5, 9, 8, 6, 4 units of work respectively, and the 5 machines have a
capacity of 11, 8, 8, 14, 9 units of work respectively.
Each job must be assigned to exactly one machine and it is not necessary to use all the 5
machines.
The cost incurred if job is scheduled on machine is given in the table below:

Table 5. Costs of assigning jobs to machines, job requirements, and machine


capacities

  Job 1 Job 2 Job 3 Job 4 Job 5 Job 6


Machine 1 3 7 5 11 9 2
Machine 2 5 8 7 1 8 1
Machine 3 4 4 8 3 7 7
Machine 4 2 3 3 6 3 8
Machine 5 1 6 9 5 4 9
50

(a) Formulate an integer programming model to determine an assignment of jobs to


machines that minimizes the total cost and does not exceed the machine capacities.
Clearly define your decision variables, objective function, and constraints.

(b) Now suppose that Job 2 and Job 5 must be scheduled on the same machine. Add
suitable constraint(s) to your formulation in part (a) to model this requirement.

(c) In addition to the assignment costs given above, suppose that a fixed cost is incurred
when machine is operated. The fixed cost of operating a machine for each of
these five machines are given by 25, 30, 24, 19 and 17, respectively. Modify the
formulation in part (a) to incorporate this fixed cost. Define any new variables, if
required.

(d) Referring to your formulation in part (c), write an inequality that models the condition
that at most 4 out of the 6 machines can be in operation.

(e) Referring to your formulation in part (c), write an inequality (or inequalities) that
models the following condition: If Machine is in operation, then at least 20% of its
capacity must be used.

(f) Referring to your formulation in part (c), write an inequality (or inequalities) that
models the following condition: If Machine 1 is operated, then at most three of the
machines can be operated.

(g) Referring to your formulation in part (c), give a constraint to model the following
condition: if at least two out of the three machines 1, 2, and 3, are in operation, then
at least one of the machines 4 or 5 must be in operation. (They don’t necessarily have
to have jobs assigned to them if they are in operation.)

Question 5
51

In this problem, you are given a branch-and-bound tree resulting from a maximization
problem defined in terms of four variables. Each of the four variables is an integer
variable, i.e., at any feasible solution, all the variables have to take on integer values. The
labels at the center of the nodes indicate the order in which the nodes were evaluated, i.e.,
its LP relaxation was solved. The notation means that a linear programming
relaxation at the node labeled k was solved and the objective function value is q. The
notation indicates that the solution to the LP relaxation at node k is
given by , , , . An active node is defined as one that still
needs to be explored and cannot be fathomed.

Figure 1. Branch-and-Bound Tree #1


52

z1LP  110.5
1
x1   2.25, 2.75, 3.5, 4.5 

x2  2
2
zLP  108.1
2
x 2   2.47, 2, 3.58, 4.42 

x1  2
3
z LP  105.2
3
x 3   2, 2, 3.41, 4.58 

x4  4
4
z LP  104
4
x 4   2, 2, 4, 4 

(a) What is the best lower bound on the optimal solution that can be deduced from
the branch-and-bound tree in Figure 1?

(b) What is the best upper bound on the optimal solution that can be deduced from
the branch-and-bound tree in Figure 1?

(c) How many nodes are still active in the branch-and-bound tree in Figure 1?
53

Figure 2. Branch-and-Bound Tree #2

z1LP  110.5
1
x1   2.25, 2.75, 3.5, 4.5 

x2  2 x2  3
2
zLP  108.1 5
zLP  110.2
2 5
x 2   2.47, 2, 3.58, 4.42  x5   2, 3, 3.41, 4.58 
x1  2 x3  3
3
z LP  105.2
3 6
x 3   2, 2, 3.41, 4.58 
6
z LP  108.7
x4  4 x 6   0.833, 4.16, 3, 5 
4
z LP  104
4
x 4   2, 2, 4, 4 

(d) What is the best lower bound on the optimal solution that can be deduced from
the branch-and-bound tree in Figure 2?

(e) What is the best upper bound on the optimal solution that can be deduced from
the branch-and-bound tree in Figure 2?

(f) How many nodes are still active in the branch-and-bound tree in Figure 2?
54

Figure 3. Branch-and-Bound Tree #3

z1LP  110.5
1
x1   2.25, 2.75, 3.5, 4.5 

x2  2 x2  3
2
zLP  108.1 5
zLP  110.2
2 5
x 2   2.47, 2, 3.58, 4.42  x5   2, 3, 3.41, 4.58 
x1  2 x3  3 x3  4
3
(g) Whatzis  105.2
LP the best lower bound on the optimal solution that can be deduced from
3 branch-and-bound tree3in Figure 3? 6 7
  2, 2, 3.41, 4.58 
xthe (1
point) 7
z LP  109
6
z LP  108.7
x4  4 x 6   0.833, 4.16, 3, 5  x 7   2, 3, 4, 4 
4
z LP  104
4
x 4   2, 2, 4, 4 

(h) What is the best upper bound on the optimal solution that can be deduced from
the branch-and-bound tree in Figure 3?

(i) How many nodes are still active in the branch-and-bound tree in Figure 3?

(j) Comment on whether the optimal solution has been found. If yes, what is the
optimum objective function value? Else, state the next step in the branch-and-bound
process.
55

Question 1
Perumal Nadar & Sons (PNS) is a manufacturer of Ayurvedic supplements prepared as
per the ancient books on Ayurveda. One of their product is Chyawanprash prepared
using sugar, ghee, honey, amla (Indian Gooseberry), and herbs. Few additional
ingredients are added based on variants of Chyawanprash. PNS is interested in finding
the optimal product mix for two variants of Chyawanprash: (a) Regular and (b) Fruit and
Nut. Ingredients such as sugar, honey and ghee are readily available in the market;
however, supply is constrained for ingredients such as ashwagandha, amla, and
Tabasheer. The quantity of ingredients required for different types of Chyawanprash is
shown in Table 1 per 1 kg of Chyawanprash. The profit earned from regular and fruit
and nut flavour per kg is 100 and 140 respectively.
Table 1 Data related to different Chyawanprash and requirements of ingredients per 1 Kg
of different Chyawanprash
Chyawanprash Availability
Regular Fruit and

Nut
Ashwagandha (in grams) 80 120 12,000
Amla (in grams) 160 140 20,000
Tabasheer (in grams) 120 180 24,000

(a) Formulate an LP model to maximize the profit for PNS. Clearly define all the
decision variables, objective function and constraints.

(b) Use the graphical method to find the optimal solution and optimal objective
function value.

(c) What is the worth of one gram of Tabasheer? Explain.

(d) What is the maximum increase in profit for Regular Chawyanprash for which the
current optimal solution remains optimal?
56

(e) Find the worth of one gram of Amla. What is the range of availability of Amla for
which this worth per gram will be valid? Show all work.

Question 2
Ajanta Garments is a fast-growing export company. The cutting and stitching sections
can each handle 100,000 man-hours per week, and the finishing section 50,000 man-
hours. The packaging section can handle 250,000 pieces per week. They produce two
types of shirts, formal and casual; two types of trousers, also formal and casual, as well as
kurtas. The contribution from these are Rs.130, 80, 150, 110 and 125 per piece,
respectively. The time required in each section is given in the table below. One of their
valued customers has an order of 5,000 sets where each set consists of 2 formal shirts and
1 formal trouser. The net contribution per set is Rs.400.

Table: Time in hours


Formal shirt Casual Formal Casual Kurta
Shirt Trouser Trouser
Cutting 1 0.5 1.5 1 1.5
Stitching 2 1.5 2 1.5 1
Finishing 0.5 0.5 0.75 0.75 0.25

Ajanta Garments wants to decide the product mix and formulates the problem as follows:

Max z = 130x1+80x2+150x3+110x4+125x5+400w
Subject to:
x1+0.5x2+1.5x3+x4+1.5x5+3.5w <= 100,000 (cutting)
2x1+1.5x2+2x3+1.5x4+1x5+6w <= 100,000 (Stitching)
0.5x1+0.5x2+0.75x3+0.75x4+0.25x5+1.75w <= 50,000 (Finishing)
w >= 5000 (Demand for set)
All variables >= 0.
The Solver output for the formulation is given below:
Variable Cells
    Final Reduced Objective Allowable Allowable
Cell Name Value Cost Coefficient Increase Decrease
$B$7 var x1 11250 0 130 120 2.857
$C$7 var x2 -2.5 80 2.5 1E+30
$D$7 var x3 -10 150 10 1E+30
$E$7 var x4 -2.5 110 2.5 1E+30
$F$7 var x5 47500 0 125 10 10
57

$G$7 var w 400 20 1E+30

Constraints
    Final Shadow Constraint Allowable Allowable
Cell Name Value Price R.H. Side Increase Decrease
$H$3 Cutting 100000 100000 22500 47500
$H$4 Stitching 100000 100000 95000 15000
$H$5 Finishing 26250 50000 1E+30 23750
$H$6 Set 5000 5000 4090.91 5000

(a) In the given formulation, explain the coefficient of w in the constraint for cutting.

(b) Find the optimal value of the decision variable w.


(c) Find the values of decision variables x2, x3 and x4 in the optimal solution.
Explain clearly.

(d) What is the optimal objective function value?

(e) Find the shadow prices of all the constraints.

(f) Ajanta Garments can outsource the cutting activity at INR 50 per hour. Is it worth
outsourcing the cutting for additional 20000 hours? What will be impact of this
on the objective function value? Explain clearly.

(g) If they increase the contribution of x1 and w by 30 and 5 respectively, and


decrease the contribution of x5 by 5, will the optimal solution change? Why or
why not?

(h) The CEO is trying to decide whether to drop the order of 5000 sets received from
the valued customer. In the long run, they are confident that they can get other
customers. What would you advise him? Give reasons.

(i) Many business analysts are suggesting that the company should move up the
value chain and make premium shirts. The contribution for premium shirt would
be 175 per shirt. Since the company does not have skilled tailors, each piece of
58

premium shirt would take 1.5 hours in the cutting section, 3 hours in the stitching
section and the same time in the finishing section. Would you advise them to go
in for the premium shirts?

(j) The CEO feels that it will take 6 months to acquire the skills provided they decide
to start making the premium shirts. Once that is done, the time taken to make the
shirts would be the same as that for the formal shirts. What would you advise
him?

Question 3
A company has one warehouse from where it supplies a product to retail stores in several
locations around the city. The demand is on 5 stretches of roads and total demand is 113
tons each day. They decide to buy their own mini trucks, and the capacity of each mini
truck is 5 tons. Each truck can make at most one trip per day.\

(i) Formulate a mixed integer program to minimize the number of trucks.


Clearly define the objective function and all the constraints.

(ii) Solve the formulation in question (i) and find the number of trucks they
should buy?

After a few days the company realized that there are delays in delivering the products.
There are many one-way streets as well. After some analysis they found that the demand
can be categorized into 5 stretches of the city. These stretches form a ring with the
warehouse in the middle. They label the stretches as 1-2, 2-3, 3-4, 4-5, and 5-1. Each
truck can service at most two stretches a day and return by the end of the day to the
warehouse. Given the one-way streets and the number of stretches a truck can do in a
day, they decide that the following 10 routes are possible, where node s refers to the start
node:
Route 1 2 3 4 5 6 7 8 9 10
number
Original s-1- s-1-2- s-2- s-2-3- s-3- s-3-4- s-4- s-4-5- s-5- s-5-1-
route 2- s 3-s 3-s 4-s 4-s 5-s 5-s 1-s 1-s 2-s
59

Thus each truck can deliver products on at most 2 contiguous stretches. For instance a
truck that leaves from the warehouse and goes to node 1, can deliver products to stretch
1-2, and 2-3 at most in a day. For ease of notation they number the routes from 1 to 10 in
the order given above. The demand on each stretch is given below.
Stretch 1-2 2-3 3-4 4-5 5-1
Demand in 25 18 17 21 32
tons

A new intern formulates the problem and gets the solution as 23 trucks with shipments as
given below.
Route 1 2 3 4 5 6 7 8 9 10
number
Original s-1- s-1-2- s-2- s-2-3- s-3- s-3-4- s-4- s-4-5- s-5- s-5-1-
route 2- s 3-s 3-s 4-s 4-s 5-s 5-s 1-s 1-s 2-s
Tons 25 18 0 17 0 0 21 0 32 0
shipped

The total shipments are met and equal 113 and there are 23 trucks which is 113/5 rounded
up. He is very happy that in spite of the restrictions on routes, the number of stretches
they can deliver on a truck per day, and the one-way streets, the number of trucks has not
gone up. However, someone looks closer at the solution and finds that 23 trucks are not
sufficient. He discovers that the number of trucks from the solution for each of the routes
with non-zero shipment is at least 5, 3.6, 3.4, 4.2, 6.4 which add up to 22.6 – which
rounded up gives 23 trucks. But in reality, Route 2 would need 4 trucks instead of 3.6.
When he added up all the trucks needed it came to 5+4+4+5+7 = 25. He was wondering
if he could do better.

(iii) Formulate a mixed integer program for the company. Clearly define all the
decision variables and constraints.

The intern is relieved when he finds that they can do with 24 trucks instead of 25 and
the solution is
Route 1 2 3 4 5 6 7 8 9 10
Tons 25 18 0 17 0 0 21 32 0 0
shipped

Here he noticed that routes 7 and 8 are contiguous and represent s-4-5-s and s-4-5-1-s
and total tons shipped is 21+32 = 53 and can be done by 11 trucks. He is a bit
relieved and shows it to the management. They say that some customers on route 8
are always served in the second half of the day (route s-4-5-1-s). Given the one way
streets they cannot ask the trucks to go in the reverse direction s-1-5-4-s. After some
60

brainstorming they decide that the tons shipped using the route s-4-5-1-s should be
restricted to 16.

(iv) State the additional constraint(s) required?

Someone then asks whether it is sufficient to minimize the number of trucks or


whether they should minimize the total cost including the cost of transportation. The
following table gives the relevant daily transport costs for each possible route.

Route 1 2 3 4 5 6 7 8 9 10
Cost 200 350 300 650 400 700 350 675 500 900
(Rs)

The cost of each truck on the road is Rs.10 lakhs.

(v) Formulate the problem to minimize the total costs. Assume they want to
recover the costs of the new trucks purchased in one year and that they operate
300 days a year.
(vi) During the monsoon they find that on one particular day the road s-1 cannot
be used. What additional constraint(s) are needed?

(vii) They find that some trucks use only one stretch in their daily travel. Such
trucks can be used for other purposes in the afternoon. If they want at least
half the trucks to be free like this, how would the formulation change?

(viii) New zoning rules on traffic restrict the daily number of mini trucks on any
given stretch to 8. What additional constraint(s) are needed?

Question 4
Consider the following integer programming (IP) problem:
Max 5x1 + 5x2
s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1, x2 integers
a) State the LP relaxation for the IP given above. Include the upper bounds on the
variable values.

b) Solve the given IP using the branch and bound (b-a-b) method (use the LR
relaxation solution provided in Appendix). Show all nodes and branches and
61

proceed as follows:

- In case both variables need branching, pick the one with the smaller value
first.
- Number the nodes in the order they are created.
- Clearly state the solution and objective function value at each node, and
whether the value obtained is a lower or upper bound for the BIP.
- Clearly state when you fathom a node and give the reason for doing so.
- State the optimal solution and objective function value obtained.

Question 5
A large part of rural India is dependent on “hand pumps” for their water needs. These work well
for water depths up to 30 meters and do not need much maintenance. Electric/diesel pumps
that work beyond this depth are not viable in remote areas since electricity and diesel might not
be available always.

Smart Pumps Co. has come up with solar pumps as a solution. For these, a high yield (at least
2000 litres/hour capacity) potable water bore well is fitted with a small solar water pumping
system and an elevated tank. The storage tank provides piped water to the doorstep of the
households. The system does away with the need for electricity, diesel and batteries, and
potentially has a long life.

The company has won a contract with the Ministry of New and Renewable Energy (MNRE) for
installation of solar pumps in various states for both irrigation and drinking water purposes. The
deliverables at the end of Jan, Feb, March and April in 2018 are 2000, 6000, 6000, and 6000
respectively. December’s ending inventory is expected to be zero.

The company has a production capacity of 5000, 4000, 8000 and 5000 pumps in Jan, Feb, March
and April 2018. Management would like a production plan that meets the following goals:

1. There are no monthly ending inventories. This is desirable since the inventory carrying
costs are high.

2. Fluctuations in monthly production levels during these months are avoided. This is
important since changing production levels attracts additional costs.

a) Formulate the production planning problem as a goal programming problem, assuming


that having no monthly ending inventories goal is of higher priority. Define all
variables clearly and state the objective function. Include goal and any other
constraints.
b) Now assume that both the goals are equally important. How would the formulation
above change?
62

Appendix: LP Relaxation Solutions


Max 5x1 + 5x2
s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 = 0.2, x2 = 2.2, objective function value = 12

Max 5x1 + 5x2


s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 <= 0
x1 = 0, x2 = 2.33, objective function value = 11.67
Max 5x1 + 5x2
s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 >= 1
x1 = 1, x2 = 1, objective function value = 10

Max 5x1 + 5x2


s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x2 <= 2
x1 = 0.33, x2 = 2, objective function value = 11.67

Max 5x1 + 5x2


s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 <= 0, x2 <= 2
x1 = 0, x2 = 2, objective function value = 10

Max 5x1 + 5x2


s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 <= 0, x2 >= 2
x1 = 0, x2 = 2.33, objective function value = 11.67

Max 5x1 + 5x2


s.t. 3x1 + 2x2 <= 5
2x1 + 3x2 <= 7
x1 <= 0, x2 <= 3
x1 = 0, x2 = 2.33, objective function value = 11.67

Max 5x1 + 5x2


63

s.t. 3x1 + 2x2 <= 5


2x1 + 3x2 <= 7
x1 <= 0, x2 >= 3 Soln: Infeasible

Question 1
The ABC Corporation makes two products with brand names ‘Swift’ and ‘Impala’. The
net contribution from each product is Rs. 1 and Rs. 2, respectively, measured in lakhs.
There are three workshops that are used to make these products. The first one called
Tuning (T) has 50 hours a week available. Each Swift requires 1 hour and each Impala 3
hours in the Tuning workshop. The second workshop is called Finisher (F) and has 40
hours available. Each product requires one hour in this workshop. The third workshop is
Assembly (A) and has 120 hours available. Swift requires 3 hours and Impala 2 hours in
the Assembly workshop.

(a) Formulate the above problem as a linear program to determine the number of Swift
and Impala products to be manufactured to maximize profit.

(b) Solve this problem graphically. Show the optimal variable values as well as the
optimum objective function value on the graph.

(c) What are the shadow prices of the constraints?

(d) What change do you expect if the T and A workshop hours are simultaneously
increased by 10 hours each?

(e) Notice that the net profit of the Swift model is half of that of the Impala model. If the
profit for Swift is further reduced by Rs. 20,000 (or Rs. 0.2 lakhs), will the company
make more Impalas and less Swifts?

(f) A new product called Gazelle is being considered for production. It yields a much
higher net profit of Rs. 3 lakhs per unit, while requiring 3,4, and 5 hours in the
Tuning, Finishing and Assembly workshops, respectively. Should the ABC
corporation take up production of the Gazelle model?

(g) Marketing says we need to produce at least as many Impalas as Swifts. What
constraint needs to be added to your formulation in part (a) and what is the new
optimal solution?

(h) The shadow price of the Tuning constraint has changed to 0.75 in the new model.
Why has it changed?
64

Question 2
Mannar & Company (M&C) is a Chennai-based sealant manufacturer, and they produce
four different types of sealants namely: (i) Dental sealants; (ii) Paint Sealants; (iii) Hot
wax; and (iv) Varnish. These sealants are made using three primary materials, namely:
(a) Silicone (S); (b) Polysulphide (PS); and (c) Silane Modified Polymer (SMP). These
ingredients are imported from Spain and due to unrest in Barcelona, M&C is
experiencing a disruption in the supply of these materials. This prompts the CEO of
M&C, Mr. Thangavelu to optimize the production plan for the month of November using
the available inventory of ingredients. The following table provides demand requirements
for various sealants during the month of November as well as the ingredients required (in
grams) to produce one kilogram of every sealant.

Sealant Quantity of ingredients required (in grams) Demand for


per kg of sealant sealants
Silicone Polysulphine SMP (in kgs)
Dental 40 25 15 Exactly 400
Paint 30 26 18 320 (maximum)
Hot Wax 60 56 26 520 (maximum)
150 (minimum)
Varnish 19 22 30
400 (maximum)
Availability of
30000 30000 21000
ingredients (in grams)

The profit per kilogram of sealants is given in the following table.


Sealant Dental Paint Hot Wax Varnish
Profit per kg 20 18 30 12

The following decision variables were used to formulate this problem as a linear
program:
Variable Description
Quantity (in kgs) of dental sealant to be produced
Quantity (in kgs) of paint sealant to be produced
Quantity (in kgs) of hot wax sealant to be produced
Quantity (in kgs) of varnish sealant to be produced

M&C has a long-term contract with Oho dental to supply exactly 400 kgs of dental
sealant every month, which cannot be violated. The following LP formulation is used for
solving the problem:
Maximize
subject to:
65

The LP model output using Excel Solver is shown below:

Decision
Final Reduced Objective Allowable Allowable
Cell Variable
Value Cost Coefficient Increase Decrease
Name
$A$2 X1 20
$B$2 X2 241.9354 0 18 0.9473 2.4257
$C$2 X3 Non-Zero 30 5.6774 1E+30
$D$2 X4 354.8387 0 12 10.56 0.6

Final Shadow Constraint Allowable Allowable


Cell Constraint
Value Price R.H. Side Increase Decrease
$A$6 Silicone 30000 0.5806 30000 1452 1400
$A$7 Polysulphine 24096.7741 30000 1E+30 5903.2258
$A$8 SMP 21000 21000 840 2292.6315
$A$9 X1 400 93.3333 47.6065
$A$10 X2 241.9354 320 1E+30 78.0645
$A$11 X3 520 1E+30 520
$A$12 X4 354.8387 150 204.8387 1E+30
$A$13 X4 354.8387 400 1E+30 45.1612

(a) How many units of dental and hot wax sealants are manufactured by M&C during the
month of November? Clearly write the reasons for your response.

(b) M&C can procure 1500 grams of silicone from the market at 500 rupees. Should they
buy these 1500 grams, and if yes, what will be the impact of this additional Silicone
on the profit?

(c) A client who procures dental sealants has shut down one of their dental clinics
resulting in a reduction of demand for dental sealant by 40 kgs (i.e., demand is now
360 kgs instead of 400 kgs). What will be the resulting impact of this reduction on the
objective function value?
66

(d) What will be the impact on profit if production of Hot Wax is increased by one unit
from its current optimal level?

(e) What will be the impact on the optimal solution if the price of paint is increased by
0.5 rupees/kg and price of varnish is increased by 4 rupees/kg.

(f) The demand for Varnish increases by at least 250 kgs. Will the optimal solution
change and will the corresponding shadow price change?

(g) A new sealant called “Acrylic Sealant” can be manufactured using 10 grams of
Silicone, 25grams of polysulphine and 40 grams of SMP per kilogram of Acrylic
Sealant. The profit earned from Acrylic Sealant is 25 rupees/kg. Should M&C make
Acrylic Sealant?

(h) What is the allowable increase and decrease for the profit per unit of Dental sealant?

Question 3
The Naturals Cosmetics Co. is considering an Ad Campaign for their new Sulphate Free
Shampoo (SFS). They are considering three TV channels for the campaign – Star,
Galaxy and Cosmos. The expected exposures on the 3 channels per Rs.1000 spent are
2200, 3000 and 1800 respectively. Creating the ads for the 3 channels costs Rs. 50,000,
Rs. 75,000 and Rs. 40000, respectively. The total budget available for the campaign after
creating the 3 ads is Rs. 280,000. Management does not want to spend more than 60% of
the budget on Galaxy. At least Rs. 60,000 should be allocated to Star and Rs. 50,000 each
to Galaxy and Cosmos. Also, the amount allocated to Star should not be more than
double the amount allocated to Cosmos.
The campaign cannot be judged as successful if the total exposures are under 750,000.
Management would like to decide the best allocation of the available advertising budget
to the three TV channels in order to maximize total exposures.

(a) Formulate a linear programming model that will help the Naturals Cosmetics Co. in
its decision. Define the variables clearly and label all constraints.

(b) Upon solving the linear programming formulation, management realizes that all the
stated conditions cannot be met simultaneously. Though the budget cannot be
increased and the minimum budget allocations for Galaxy and Cosmos are
67

mandatory, management is willing to relax the other conditions. They would like to
achieve the following in the given order of priority:

1. Avoid allocating to Star, more than double the amount allocated to Cosmos.
2. Avoid allocating less than Rs. 60,000 to Star.
3. Minimize the under achievement of the goal of 750,000 exposures.
4. Avoid expenditure of more than 60% of the available money on Galaxy.

State the goal programming problem that needs to be solved next. Clearly define all
variables used and label all constraints.

(c) Management is considering the following decision alternatives:

Decision 1 Decision 2 Decision 3


Funds allocated to
60 0 62
Star
Funds allocated to
170 230 168
Galaxy
Funds allocated to
50 50 50
Cosmos

Determine the amount of over-achievement and under-achievement associated with


each goal, for each of the three decisions under consideration.

(d) If management were to rank the three decision alternatives from most preferred to
least preferred, consistent with their stated goals, what would this ranking be?
Explain why?

Question 4
The year is 2030. Alka Yagnik is making a comeback. You are a high-powered audio-
visual studio executive and your current task is to decide what songs to include on two
new albums of Alka Yagnik that your recording company is slated to produce. Note that a
song, if selected, must be added in its entirety to an album. Alka has recorded two types
of songs that we will categorize as “romantic” tunes and “dance” tunes. The details of all
the recorded songs are given in the table below.

S. No Type of Song Duration (mins)


1 Romantic 5.01
2 Romantic 7.43
3 Dance 6.49
68

4 Romantic 7.71
5 Dance 6.85
6 Dance 7.33
7 Dance 6.84
8 Romantic 6.88
9 Dance 5.05
10 Dance 5.84
11 Dance 6.54
12 Romantic 6.98
13 Romantic 6.85
14 Romantic 7.25
15 Romantic 6.29

Moreover, the songs selected to appear on the two albums must satisfy the following
conditions:

1. A song can appear on at most one album


2. Duration of Album #1 can be at most 40 minutes
3. Duration of Album #2 can be at most 35 minutes
4. Each album must have at least two and at most four romantic tunes
5. Album #1 must have at least three dance tunes

(a) Formulate an integer program that will determine a feasible allocation of Alka
Yagnik’s songs to each album such that the total duration of songs in the two albums
is maximized.

(b) Using your formulation in part (a), determine any feasible allocation of songs to
albums. Your allocation must be such that it should not be possible to add any further
songs to either Album 1 or Album 2. If the optimum total duration of songs in both
albums is 74.80 minutes, what is the percentage deviation of your (feasible) objective
function value from optimality?

(c) Now suppose that you wish to model the condition that either Song 6 or Song 10 but
not both can be in Album #1. What constraints would you add to part (a) to model
this restriction?

(d) Now suppose that if either Song 12 or Song 13 (or both) is included in Album #1,
then Song 15 must not be included in both albums. Add appropriate constraint(s) to
part (a) to satisfy this restriction.
69

(e) Now suppose that you wish to model the restriction that if at least two out of the three
songs, namely Songs 9, 10, 11, are on Album #1, then at most one of the two songs,
Song 5 and Song 6, must be on Album #2. What constraints would you add to your
model in part (a) to account for this restriction?
70

Question 5
Consider the following 0-1 integer programming problem:

Maximize
subject to:

You are given a partially completed branch-and-bound tree below, where the labels at the
center of the nodes indicate the order in which the nodes were evaluated, i.e., its LP
relaxation was solved. The notation indicates that a linear programming
relaxation at the node labeled k was solved and the objective function value is q. The
notation indicates that the solution to the LP relaxation at node k is
given by , , , . An active node is defined as one that still
needs to be explored and cannot be fathomed.
71
72

You are required to complete the branch-and-bound tree and determine the optimal
solution to the problem. Towards this end, answer the following questions.

(i) What is the current incumbent (or candidate) solution?

(j) What is the best upper bound on the problem at this stage?

(k) What are the active nodes in the tree at this stage?

(l) Formulate the linear programming relaxation to be solved at Node 5.

(m) Graphically represent the formulation from part (d) and determine the optimal
solution at this node and corresponding objective function value. Clearly indicate the
corner points of the feasible region, the feasible integer points, and optimal solution.

(n) Based on your solution to part (d), did the incumbent value change? If yes, what is
the new incumbent solution?

(o) What are the active nodes in the branch-and-bound tree at this stage?

(p) From the solution shown at Node 6, select an appropriate branching variable and
determine the optimal solution at Node 7. Did your incumbent solution change?

Determine the optimal solution at Node 8. Can the branch-and-bound process be


terminated at Node 8? If yes, explain your reasoning. If not, continue branching and
determine the optimal solution to the problem.

You might also like