Package teamwork :: Package math :: Module ProbabilityTree

[frames] | no frames]

Module ProbabilityTree

source code

Defines the layer of probabilistic branches over KeyedTree

Classes

[hide private]

ProbabilityTree
A decision tree that supports probabilistic branches

Functions

[hide private]

createBranchTree(plane, falseTree, trueTree)
Shorthand for constructing a decision tree with a single branch in it

source code

createNodeTree(node=None)
Shorthand for constructing a leaf node with the given value

source code

createEqualTree(plane, equalTree, unequalTree)
Shorthand for constructing a decision tree that branches on whether the value lies on the plane or not, with the former/latter cases leading down to the given equalTree/unequalTree

source code

createDynamicNode(feature, weights)
Shorthand for constructing a leaf node with a dynamics matrix for the given key with the specified weights (either KeyedVector, or just plain old dictionary, for the lazy)

source code

ProbabilityTree

createANDTree(keyWeights, falseTree, trueTree)
To create a tree that follows the True branch iff both the actor has accepted and the negotiation is not terminated: source code

ProbabilityTree

createORTree(keyWeights, falseTree, trueTree)
To create a tree that follows the True branch iff either the actor has accepted or the negotiation is not terminated: source code

ProbabilityTree

identityTree(feature)
Creates a decision tree that will leave the given feature unchanged

source code

Variables

[hide private]

EMPTY_NAMESPACE = None

EMPTY_PREFIX = None

StringTypes = (<type 'str'>, <type 'unicode'>)

XMLNS_NAMESPACE = 'http://www.w3.org/2000/xmlns/'

Function Details

[hide private]

createBranchTree(plane, falseTree, trueTree)

source code

Shorthand for constructing a decision tree with a single branch in it

Parameters:

plane (Hyperplane) - the plane to branch on
falseTree - the tree that will be followed if the plane tests False
trueTree - the tree that will be followed if the plane tests True
falseTree,trueTree (ProbabilityTree)

Note: Will not prune tree

createANDTree(keyWeights, falseTree, trueTree)

source code

To create a tree that follows the True branch iff both the actor has accepted and the negotiation is not terminated:

>>> tree = createANDTree([(StateKey({'entity':'actor','feature':'accepted'}),True), (StateKey({'entity':'self','feature':'terminated'}),False)], falseTree, trueTree)

Parameters:

keyWeights ((Key,boolean)[]) - a list of tuples, (key,True/False), of the preconditions for the test to be true
falseTree - the tree to invoke if the conjunction evaluates to False
trueTree - the tree to invoke if the conjunction evaluates to True
falseTree,trueTree (DecisionTree)

Returns: ProbabilityTree

the new tree with the conjunction test at the root

Note: the default truth value of the plane is True (i.e., if no keys are provided, then trueTree is returned

createORTree(keyWeights, falseTree, trueTree)

source code

To create a tree that follows the True branch iff either the actor has accepted or the negotiation is not terminated:

>>> tree = createORTree([(StateKey({'entity':'actor','feature':'accepted'}),True), (StateKey({'entity':'self','feature':'terminated'}),False)], falseTree, trueTree)

Parameters:

keyWeights ((Key,boolean)[]) - a list of tuples, (key,True/False), of the preconditions for the test to be true
falseTree - the tree to invoke if the conjunction evaluates to False
trueTree - the tree to invoke if the conjunction evaluates to True
falseTree,trueTree (DecisionTree)

Returns: ProbabilityTree

the new tree with the conjunction test at the root

Note: the default truth value of the plane is False (i.e., if no keys are provided, then falseTree is returned

identityTree(feature)

source code

Creates a decision tree that will leave the given feature unchanged

Parameters:

feature (str/Key) - the state feature whose dynamics we are creating

Returns: ProbabilityTree