SAMBA: Data Analytics Imperial College Business School

Mock Final Exam

Total Marks Available: 100

Instructions

1. This is a closed-book exam.

2. All answers and explanations to your answers (including computational work) must be pro-
vided. Correct answers without explanations cannot get full marks.

3. Please report your answer and explanation for each question in the answer box.
You can upload your Excel work at the end of the exam.

4. Keep your answers succinct and to the point.

5. Please download Data.xlsx before you start. The file contains the data in Question 2.

Advice: Read the entire exam carefully before starting on your answers.

1
Question 1. (20 marks)

(a) The Broadmoor Hotel in Colorado Springs contains 700 rooms and is on the Gold List
Condé Nast Traveler in 2004. Suppose Broadmoor’s marketing group forecasts a mean
demand of 670 rooms for the upcoming weekend. Assume that demand for the upcoming
weekend is normally distributed with a standard deviation of 35. What is the probability
that all the hotel’s rooms will be rented in the upcoming weekend? (10 marks)

(b) Yearly MOT (Ministry of Transport) tests are required for vehicles over three years
old in the UK. The tests check whether vehicles meet road safety and environmental
standard. Suppose that 15% of all inspected vehicles in the UK have problems that need
to be corrected. Unfortunately, MOT tests fail to detect these problems 8% of the time.
Consider a car that gets an MOT test and is found to be free of problems. What is
the probability that there is indeed something wrong that the MOT test has failed to
uncover? (Note: A car that actually has not problem will always be found to be free of
problems in MOT tests.) (10 marks)

2
Question 2. (30 marks)

A manager of boiler drums wants to predict the number of worker-hours required to erect
boiler drums in future projects using regression analysis. Data for several randomly selected
boilers have been collected in the file Data.xlsx. The dataset includes the dependent variable
Worker Hours and the following independent variables:

• Boiler Capacity: measured in pounds per hour.

• Design Pressure: measured in pounds per square inch.
• Boiler Type: a binary variable, 0 = industrial, 1 = utility.
• Drum Type: a binary variable: 0 = mud, 1 = steam.

(a) Estimate a multiple regression equation to predict the number of worker-hours needed to
erect boiler drums using all four independent variables. Interpret the coefficient estimate
for Design Pressure and its statistical significance (use 0.05 as the significance level). (10
marks)
(b) According to the estimated regression equation, what is the difference between the mean
number of worker-hours required for erecting industrial and utility field boilers, assuming
all the other independent variables are held constant? (5 marks)
(c) Given the estimated regression equation, predict the number of worker-hours needed to
erect a utility-field, steam-drum boiler with a capacity of 550,000 pounds per hour and
a design pressure of 1,400 pounds per square inch. (5 marks)
(d) Add the square of Design Pressure to the regression equation and estimate the regression
equation again. Report the new R2 and interpret its meaning. (5 marks)
(e) Given the estimated regression equation in part (d), do you think whether there is
evidence of a non-linear relationship between Worker Hours and Design Pressure? Why
or why not? (5 marks)

3
Question 3. (20 marks)

Bob Farrell, an architect, is considering buying, restoring, and reselling a house. The cost
of the house is $240,000 and Bob believes it can be sold for $450,000 after being restored.
Bob expects that he can sell the house as soon as the restoration is completed. He expects
to pay $1500 a month in finance charges from the time he purchases the house until it is
sold. Bob has developed two sets of plans for the restoration. Plan A will cost $125,000
and will require three months to complete. This plan does not require changes to the front
of the house. Plan B is expected to cost $85,000 and require four months of work. This
plan does involve changes to the front of the house, which will require the approval of the
town’s historic preservation committee. Bob expects the approval process for plan B to
take two months and cost about $5,000. Bob thinks there is a 40% chance that the historic
preservation committee will approve this design.

Bob plans to buy the house immediately but cannot decide what he should do next. He could
proceed immediately with restoration plan A. Alternatively, he could also start immediately
with restoration plan B. If he starts immediately with plan B, he can first work on some
other parts in his plan while waiting for the decision of the historic preservation committee
on changes to the front of the house. As mentioned, he will not know the decision for two
months. If the committee approve it, Bob may continue working on plan B and finish it in
another two months (i.e., 4 months in total from starting with plan B to finishing it). If
the committee do not approve it, he will have to start over and implement plan A instead.
Bob estimates that starting over with plan A would cost an additional $10,000 over plan
A’s normal cost due to the cost of partial implementation of plan B first. If he starts over
with plan A, he still needs another three months to complete plan A (i.e., 5 months in total
from starting with plan B to finishing restoration with plan A). A third option for Bob is
to hold off implementing either plan until he knows the outcome of the historic planning
committee’s decision.

1. Construct a decision tree to model Bob’s decision problem. Your tree should clearly
display decision nodes, event nodes and terminal nodes as well as any cash flows and
probabilities associated with the nodes and branches. Possible actions and outcomes
should be clearly labelled on the tree. At each terminal node in the tree you should
include the total cash flow for the path leading to that terminal node. You do not need
to solve the tree in this part. (10 marks)

2. If Bob chooses to start with plan B immediately, what is his expected payoff? Explain
your answer. (5 marks)

3. What is the optimal decision for Bob right now (i.e., to start with plan A, start with
plan B, or hold off)? Explain your choice. (5 marks)

4
Question 4. (30 marks)

Valu-Com Electronics manufactures five different models of telecommunications interface

cards for personal and laptop computers. As summarized in the following table, each unit
of these devices requires differing amounts of printed circuit (PC) board, resistors, memory
chips, and assembly.
Hyperlink Fastlink Speedlink Microlink Etherlink
PC Board (square inches) 20 15 10 8 5
Resistors 28 24 18 12 16
Memory Chips 8 8 4 4 6
Assembly Labour (in hours) 0.75 0.6 0.5 0.65 1

The unit profit for each model are as follows.

Hyperlink Fastlink Speedlink Microlink Etherlink
Per Unit Profit ($) 53 48 33 32 38

In their next production period, Valu-Com has 80,000 square inches of PC board, 100,000
resistors, 30,000 memory chips, and 5,000 hours of assembly time available. The company
can sell all the product it can manufacture, but the marketing department wants to be sure
that it produces at least 500 units of each product and at least twice as many FastLink cards
as HyperLink cards while maximizing profit.

(a) Formulate a linear programming model for this problem. Be sure to clearly specify your
decision variables, objective function, and constraints. (15 marks)

(b) Solve the model using Excel Solver. What is the optimal production plan for Valu-
Com? How much is the optimal total procurement cost? (10 marks)

(c) Suppose that Valu-Com Electronics implements the optimal production plan. In this
case, consider the possibility of the company scheduling its assembly workers for over-
time. Could this decision potentially increase Valu-Com’s profit? Explain your answer.
(5 marks)