Decision trees

Question 1

Define decision trees.

Answer 1

A business can’t afford to follow every option so it may use a decision tree to analyse the probability of a success in a choice of strategies;

• A new product launch
• A new marketing campaign on the current product
• Relocation to a new building

Question 2

How do decision trees work?

Answer 2

A decision tree traces an alternative outcome of a decision, the results can be compared and the business can decide which option would be the most profitable

This is a quantitative approach (means it uses numbers) it is pictorial as it’s in the shape of a tree and it predicts the best bet outcome so a business knows what to spend its money on

The data is usually based on historical data from the business, for example they would know the probability of success of launching a new product would be 20% based on previous product launches and how well they had turned out

Question 3

How do you draw a decision tree?

Answer 3

Your start with a decision point which is a square.

Branches with different options come out of this square. The branches have chance nodes (circles) on the end which show different outcomes.

Chance nodes have success and failure probability data coming out of them.

Probability of a node always adds up to 1.

You can add expected profit or loss at the end of probability

Expected value can be calculated by multiplying probability of success or failure into the expected profit of of it.

Question 4

What are the limitations of decision trees?

Answer 4

Decision trees are based on predicted data of the potential impact of a decision

Decision tree does not take into account unforeseen costs and circumstances
Probabilities and net outcomes are all estimates

Using decision trees might fit with the culture exhibited by the business but this is likely to be overridden by the need to act quickly, which would suggest the decision trees are not that useful in some situations

Large degree of uncertainty about any situation that a decision tree might be used for, so the benefits of adding numbers to uncertainty e.g. 58% might not be correct