GeNIe Tutorials: Tutorial 12 - Sensitivity analysis
From DSL
Objective of the tutorial: To learn how to perform Sensitivity Analysis in GeNIe.
Estimated Time: 15 Minutes
At the end of the tutorial you will be able to:
- Add a Sensitivity node to the network for Sensitivity analysis.
- View results that have a multidimensional table.
GeNIe supports simple sensitivity analysis in graphical models. To perform sensitivity analysis, add an additional indexing variable that will index various values for parameters in question and have GeNIe compute the impact of these values on the results. We will demonstrate this idea on the example diagram introduced in the tutorial section on influence diagrams.
Open the Bayesian network created in the Creating Influence Diagrams tutorial.
If you do not have it saved, you can find a copy in the GeNIe/Example Networks folder. It is named tutorial4.xdsl.
You should have the following network loaded in Graph View:
Suppose that we are uncertain as to the actual probability of the success of the venture. Believing that the nominal value of 0.2 is approximately right, we feel that it can be as low as 0.1 and as high as 0.35. To express this we will add a Decision node called Sensitivity with three states: Low, Nominal, and High.
Create a Decision node by selecting the 
(Decision) tool from the Tool menu or the Standard toolbar and click on some empty space near the network.
Name it as Sensitivity.
We need 3 states for this node for modelling the 3 sensitivity levels we have.
Define 3 states for the Sensitivity node, name them as Low, Nominal and High.
As this Sensitivity is regarding the probability of Success, we need to define a relationship between the Sensitivity node and the Success of venture node.
Add a directed arc from Sensitivity node to the Success node.
We may add an arc from Sensitivity node to Investment Decision node in order to introduce an explicit temporal order between the decisions. This arc will be dotted to indicate that it signifies temporal order between nodes.
Add a directed arc from Sensitivity node to the Investment Decision node.
If you forget to add this arc, don't worry, because GeNIe will automatically assume the temporal order of Sensitivity node before Investment Decision node and draw the arc for you.
The states of node Sensitivity will index the parameters in question and will allow to specify their low, nominal, and high values. We enter the low, nominal, and high values for the probability of outcome Success in the conditional probability table of the node Success.
Define the probabilities for each level of sensitivity for the Success of venture node as shown below:
Now we are ready to update the model and observe the result.
Update the model.
Click on Value tab of Node Properties Sheet of Investment Decision node.
You will see the following result:
We can see that even the most optimistic value of the probability of success does not make investment an attractive option, so our decision is not sensitive to the value of the probability of success.
We can also observe the impact of uncertainty over a parameter on the posterior probability distribution of any node in the network. In the example above, we can examine the posterior probability of the node Expert Forecast and see its distribution depending on our initial estimates of the probability of success.
The modified influence diagram for Sensitivity Analysis is saved as Sensitivity.xdsl in the Example Networks folder.
This concludes the tutorial on Sensitivity Analysis in GeNIe.
Next Tutorial:
