In project risk analysis, Decision Trees are a valuable tool as they clearly outline to decisions that are being evaluated and the results are easy to interpret. Let’s take a look at a simple example. .A group of gangsters are planning to rob a casino in Las Vegas. They have three choices:
- August Place casino: will yield $2M from the vault or but nothing if they cannot crack it. They estimate that probability that they can break into the vault equals 40%.
- Washington-Washington casino: will yield $1M.
- Illusion Casino: will yield $2M, if they get help from an insider (30% probability) and $1M without inside help.
We will plot this analysis on a decision tree. The decision tree has three different types of nodes:
- Decision nodes: indicates a juncture where a decision must be made on one branch or another
- Uncertainty nodes: indicates that branches have probabilities associated with them
- End nodes: indicates an outcome of the decision
Calculation of decision trees start from left to right. For each chance node we need to calculate expected value. In our case, an expected value is the expected reward from a specific robbery alternative. It is an indicator that we are going to use to make choose the option with the largest expected value. Expected value (EV) is calculated as probability multiplied by the outcome. For the Casino 1 alternative the EV is $2M * 40% + $0M * 60% = $0.8M. For Casino 2 the EV is $1M as the probability equals 100%. For Casino 3, the EV is $2M * 30% + $1M * 70% = $1.3M.
Using expected value as their criteria, the decision for bank robbers is now straight forward. Casino 3 has highest expected value even taking into account the uncertainty associated with the possibility that they will not receive any insider help. This is indicated by the arrow pointing from decision node to the Casino 3 branch.
During project planning, decision trees can use notional (simplified) schedules to generate decision trees. To do this create project schedules for each alternative (branch). Alternative scenarios can be represented by different parallel paths through the schedule. Different paths of a project schedule are the result of branching when a predecessor activity has more than one successor activity. The end nodes of the decision tree will represent cost and/or duration of each alternative. It is important to include risks and uncertainties as part of this process as they can obviously have a significant impact on the expected values.
Decision Trees can be a useful tool to incorporate in your project risk analysis and decision making, By calculating expected values and considering uncertainties, Decision Trees help you to make informed choices that maximize benefits and minimize risks. In our next blog, we will discuss how to incorporate branching into your project decision analysis processes.