bbn.sensitivity() conducts a sensitivity analysis on a Bayesian Belief Network (BBN) model.
It evaluates the impact of varying key node on the network's outcomes using bootstrapping.
The analysis helps identify which node significantly influence the network, providing insights into the robustness and dependency of the network's structure.
bbn.sensitivity(bbn.model, boot_max = 1000, ...)The function outputs a plot showing the nodes most influential to the network's outcomes, alongside a table ranking these variables by their impact.
The analysis highlights how changes in the key nodes can affect the network, offering valuable insights for model refinement and decision-making.
a matrix or dataframe of interactions between different model nodes.
One or more nodes (recommended no more than 3) which would be the main outcomes of interest in the model.
The spelling of these nodes needs to be identical (including capital letters) to that in the matrix or dataframe file.
(note, you should include spaces if these are in your matrix or dataframe file, rather than the dot notation used once imported into R).
The number of bootstraps to perform. Suggested range for exploratory analysis 100-1000. For final analysis recommended size = 1000 - 10000 - note, this can take a long time to run. Default value is 1000.
Key nodes for sensitivity analysis.
The function is designed to handle up to three key nodes, beyond which it recommends limiting the analysis for clarity and efficiency.
data(my_BBN)
bbn.sensitivity(bbn.model = my_BBN, boot_max = 100, 'Limpet', 'Green Algae')
Run the code above in your browser using DataLab