 Choosing a Tornado Diagram

To run a Tornado Diagram, choose the appropriate Tornado type from the Tornado drop down menu in Home | Sensitivity. Note that each type of tornado diagram has advantages and disadvantages. You will often want to use more than one in the course of an analysis.

Most useful with deterministic models to help determine which variables should be modeled probabilistically.

Pros:
• Shows a variable's impact on the full model (probabilistic model if there are chance events in the model).
Cons:
• Can't compare chance events with values.
• Requires two runs of full model for each bar.

Most useful with large probabilistic models as a quick method of running a sensitivity analysis, and also has the benefit of being easy to explain to others in presentations.

The Initial Decision Alternatives Tornado Diagram displays several Base Case Tornado Diagrams on one chart (one tornado for each alternative of the initial decision). Therefore, the pros and cons of the Base Case Tornado Diagram also apply to the Initial Decision Alternatives Tornado Diagram.

Pros:

• Can compare chance events and values.
• Fastest to run.
Cons:
• Ignores probabilities.
• Removes conditioning and testing.

Useful because the results can be easily compared to other results such as a Policy Tree or Risk Profile from the model. Also has the benefit of being fairly easy to explain to others in presentations.

Pros:

• Can compare chance events and values.
• The base run (vertical line in the tornado) will equal the expected value of the model as in a Risk Profile or a Policy Tree
Cons:
• Takes longer to run

Determinstic Event Tornado Diagrams are most useful with probabilistic models where it is important to incorporate probabilities into the sensitivity analysis (e.g. the probability distributions vary widely among chance nodes or many chance nodes only condition the probabilities of other chance nodes). You should use a Probabilistic Event Tornado Diagrams if your model incorporates Risk Tolerance.

Pros:

• Shows contribution to risk
• Uses expected value
Cons:
• Can't compare chance events with values
• May take a long time to run
• Hardest to explain

Versions: DPL Professional, DPL Enterprise, DPL Portfolio