Business Analytics Decision Making Under Uncertainty

Business Analytics Decision Making Under Uncertainty

• Business analytics involves decision-making in uncertain situations.
• A formal framework analyzes decision problems with uncertainty.
• This includes criteria to choose between alternatives, how probabilities are used in the decision-making process, and how earlier decisions affect later ones.
• It also details quantifying information value and how risk attitudes influence analysis.
• A decision tree is a powerful tool for visualizing all important aspects of a problem.
• Decision trees present the decision alternatives, uncertain outcomes, probabilities, economic consequences, and chronological events.

Elements of Decision Analysis

• Decision analysis problems have common elements:
  • Identifying the problem needing a solution.
  • Defining possible decisions.
  • Determining possible outcomes with uncertainty.
  • Assessing probabilities of outcomes occurring.
  • Determining payoffs or costs for each decision and outcome combination.
  • Choosing a decision criterion for determining the best decision.

Identifying the Problem

• When a trigger initiates the need to solve a problem, the problem itself must be explicitly defined.
• Potential decisions depend on how the problem is specified.
• Deciding before all uncertain results are known is a common challenge in decision making under uncertainty.

Probabilities of Outcomes

• Accurately assessing probabilities of outcomes is challenging and often subjective.
• Historical data may contribute to understanding possible outcomes.
• Subjective estimates are necessary when new products are introduced.
• Probabilities can change as further information becomes available.

Payoffs and Costs

• Decisions and outcomes have consequences, which can be monetary or non-monetary.

Decision Criteria

• Different criteria can be used for making decisions, including looking at the best or worst possible outcome, the 5th percentile of the outcome distribution, variance of outcomes, downside risk, and expected monetary value.

• Expected monetary value (EMV) is a weighted average of possible payoffs, considering probabilities.

• It's the most commonly used criterion for decision problems.

More About EMV Criteria

• EMV values can be used like a sure monetary outcome for decisions.
• Calculating EMV using spreadsheet software is straightforward, enabling sensitivity analysis of input variables.
• EMV values are calculated using SUMPRODUCT(Array1,Array2)

EMV and Decision Trees

• Decision trees visually represent decision problems, including events, decisions, and probabilities, and monetary values.
• Decision nodes are presented as squares, representing a time where a decision is made.
• Probability nodes are presented as circles, representing the time period where an outcome is uncertain.
• Time in the decision tree proceeds from left to right. Any branches leading out of a node are not yet occurred.

Branches and Nodes in a Decision Tree

• Branches from decision nodes represent possible decisions.
• Branches from probability nodes represent possible outcomes.
• Probabilities are usually conditional on observed events.
• Monetary values are usually indicated at the end nodes of the decision tree.

Folding-Back Procedure in Decision Tree

• To use a decision tree to find EMVs and optimal decisions, the folding-back procedure starts at the end nodes and works left, determining a maximum for the EMV decision.
• At each chance node, calculate an EMV of the sum of products of monetary values and probabilities of each node.

One-Stage Decision Problems

• In single-stage decision problems, the decision is made immediately.

Example 6.1: New Product Decisions at Acme

• Goal: Use EMV to help Acme decide if marketing a new product is worthwhile.
• Acme's financial inputs are estimated fixed cost and unit margins.
• The sales volume is uncertain, and different possibilities (Great, Fair, Awful) are considered.
• EMV calculation involves multiplying each sales volume by the unit margin and subtracting fixed costs.

Example 6.2: New Product Decisions with Technological Uncertainty

• Goal: Use a decision tree to determine Acme’s best strategy for a new product with technological uncertainty.
• Acme can stop development at any point.
• The probability of technological failure is a key factor.
• The optimal path through the decision tree using the TRUE branches can be considered to identify the best strategy.
• Calculating the EMV gives the company a numerical value of the strategy’s worth.

Example 6.3: New Product Decisions with Options to Buy Information

• Goal: To decide if it is worthwhile to purchase marketing research information.
• Acme can decide whether to buy information.
• Using conditional probabilities, Acme can decide if marketing the product is worthwhile.

Bayes’ Rule

• Bayes’ rule is a formal mathematical mechanism for updating probabilities as new information becomes available.
• Prior probabilities are initial probabilities, and likelihoods are conditional probabilities

Sensitivity Analysis

• A strategy region graph displays how the EMV varies with input variable conditions, demonstrating whether the optimal decision changes in response to input changes.
• A two-way sensitivity chart shows how the EMV varies when multiple input variables change simultaneously.

This quiz explores the key concepts in business analytics related to decision-making under uncertainty. It covers decision analysis frameworks, tools like decision trees, and the factors that influence risk and outcomes. Test your understanding of how to analyze decision problems and apply probability in decision processes.