Decision Analysis

Problem:  Wildcat Industries is a small oil speculation company based in Alberta. They purchase tracts of land that larger oil companies dismiss as unpromising even though initial geological reports suggest may have oil. Then Wildcat does more sophisticated seismic analysis to determine if they should attempt to drill for oil. If oil is found, they can sell the land to one of the larger companies for a huge profit or they can extract the oil themselves. Wildcat recently bought various tracts of land in Manitoba. A geophysicist recently completed their seismic analysis, and they say there is a one in three chance that one tract of land has oil.

Wildcat Industries must now decide what to do. They can drill for oil at a cost of $250,000. If land is dry (i.e. no oil), then they lose the entire investment. But if a well is found, they can sell the land for a large profit, or they extract the oil themselves. If a large oil well is found, they’ll sell the property for a projected $1.5 million. If a small oil well is found, they plan to extract the oil quickly themselves at a cost of $25 per barrel and sell the oil for $45 per barrel. A small oil well has around 50,000 barrels. If there is oil, there is a 50% chance it is a small well.

Alternatively, another oil speculator has learned about the geophysicists report and is offering to buy the land for $90,000.

Decision Analysis

For your assigned question, you are asked to make a decision in an uncertain situation based on multiple decision methods.

Step 1: Read the problem

Step 2: Create payoff and opportunity loss tables for the problem.

Step 3: Determine the optimal decision based on all of the decision methods:

• Maximax
• Maximin
• Expected monetary value
• Expected opportunity loss
• Return to risk ratio

Step 4: Based on your work in Step 3, decide what you believe is the optimal decision and explain your reasoning.

Step 5: Create a report of your work for your manager. The report needs to include the following:

• A clear statement of the suggested decision and the reasoning behind your decision. Your reasoning must rely on the decision methods from Step 3 (that is, your personal opinion is not appropriate) and an explanation of why you valued certain decision methods over others.
  • Make sure to talk about multiple decision methods and not just focus on one. That is, don’t just say “EMV is the best”. Instead, explain why EMV is the best and include what the other decision methods indicate as well.
  • This report should be a maximum of 1 page, single spaced long, 12 pt Times New Roman long.
• At the END of your report, include three tables that are properly labeled with these names
  • Table 1: Payoff Table
  • Table 2: Opportunity Loss Table
  • Table 3: Summary of results of decision analysis

Table 3 would looks something like this but with “Choice 1” changed to the actual choice in the problem.

Two very important notes:

1. This is a statistics class and the goal of this assignment is to demonstrate your understanding of regression and making decisions under uncertainty. When you are reviewing your work, ask yourselves “are we demonstrating our understanding of those topics?”
2. Related to 1, though the rubric is in the middle of the document, it is the most important part of the assignment as it specifically tells you what you are being grade on. As you complete each step, ensure that you have checked your work against the rubric to make sure you are maximizing your grade. Also it indicates where to put most of your effort (i.e. the portion of the assignment that is worth the most should be where you put most of your work).

Learning objectives:

Through this assignment, demonstrate that you have obtained the following learning objectives:

• Read and interpret output of regression analysis in Excel to determine which variable is the “best” predictor.
• Utilize multiple decision methods to analyze a situation and arrive at an optimal decision.

Hints and suggestions for report writing

General overview of the report.

• Start with a one sentence overview of the problem. For example, “We were tasked with determining the best variable to predict the number of labour hours for moving.”
• Then clearly state the answer to the problem. For example, “We determined that number of large objects moved was the best predictor of labour hours.”
• Then break the remainder of the report into sections that have well-named headers.
  • For example, the section that explains why large objects is the best predictor, could be called “Why large objects is the best predictor”.
  • Headers should stand out in the page (note the headers in this document).
  • Everything written within that section needs to relate to the heading. For example, in the section “Why large objects is the best predictor”, don’t start talking about the interpretation as that is not relevant to that section.
  • Choose an order of the sections based on what makes sense for your argument. That is, it does not need to follow the order of the outline of the report in Step 4 for regression and Step 5 for making decisions.

Suggestions:

• The report is being written for a manager who likely knows statistics. Do not include definitions, explanations of your calculations and such. The manager knows how to do these things and doesn’t need you to reexplain them.
• Most managers are busy so make sure your report is succinct, to the point and clear.
• For regression, you can include a scatterplot. If you do call it: “Figure 1”.
• For both reports, you can refer to the tables and figures by name. For example, “As we see in Table 3, the most common optimal decision is building a small factory.”
• Write the report. Then leave it for two days. Read it again with fresh eyes. This helps ensure the document is clearly written and can aid in finding typos.

Breakdown of marks:

Here is the breakdown of your mark. For your reference, all/most/some/few/none (colour coding matches for different words) are defined as follows.

• All: Either everything is done correctly or there are only a very small number of errors/omissions.
• Most: Either there are several small errors/omissions or there is one significant error.
• Some: There is more than one significant error but overall the answer is right.
• Few: There is more than one significant error and overall the answer is not quite right.
• None: Either the component is missing, or it is included but it is very wrong.

Table 1 – Breakdown of marks