Command:
setCorrelationTree ( |
|
What:
Sets the correlation dependence-tree method as method to be used.
Parameters:
numCorr [integer] |
the number of correlation parameters |
corrChild [array of integer] |
a preallocated array with size numCorr to store the children |
corrParent [array of integer] |
a preallocated array with size numCorr to store the parents |
corrValue [array of double] |
a preallocated array with size numCorr to store the correlation values |
Return:
error_code [int]: 1 if the operation has occurred without errors, 0 if an error has occurred
Examples:%tree 1 - 2 - 4
% \ \
% 3 5
% 3 5
children=[2,3,4,5]
parent=[1,1,2,2]
value=[-0.8,0.2,0.9,-0.6]
setCorrelationTree(4,children,parent,value)
setCorrelationTree(4,children,parent,value)
this command in Matlab sets the correlation Tree method. Note that factors 2 and 3 are children of factor 1, while factors 4 and 5 are children of factor 2.
About the correlation method:
The dependence-tree method for
modeling the correlation between factors (see [9] and [10]) is related
to influence diagrams, but makes use of undirected acyclic graphs
instead of directed graphs used in influence diagrams.
The user can specify correlations among input factors that form a tree structure. Whatever correlation values are imposed by the user in this way, it is guaranteed that a joint pdf exists. The joint pdf has minimum information amongst all those joint distributions which satisfy the criteria given by the user.
The user can specify correlations among input factors that form a tree structure. Whatever correlation values are imposed by the user in this way, it is guaranteed that a joint pdf exists. The joint pdf has minimum information amongst all those joint distributions which satisfy the criteria given by the user.