starvine.vine.C_vine.Ctree Class Reference

A C-tree is a tree with a single root node. More...

Inheritance diagram for starvine.vine.C_vine.Ctree:

Public Member Functions
def	__init__ (self, data, lvl=None, kwargs)
	A single tree within vine. More...
def	seqCopulaFit (self)
	Iterate through all edges in tree, fit copula models at each edge. More...
def	treeNLLH (self, treeCopulaParams=None)
	Compute this tree's negative log likelyhood. More...
def	evalH (self)
	Define nodes of the T+1 level tree. More...
Public Member Functions inherited from starvine.vine.tree.Vtree
def	__init__ (self, data, lvl, parentTree=None, kwargs)
	A generic tree within vine. More...
def	addNode (self, dataLabel, data)
	Add a node to the tree. More...
def	buildNodes (self)
	Assign each data column to a networkx node.
def	setEdges (self, nodePairs=None)
	Sets the node to node connections in the tree. More...
def	lowerTree (self, lowerTree=None)
	Gets lower tree. More...
def	lowerTree (self, lTree)
	Sets lower tree.
def	upperTree (self, upperTree=None)
	Gets upper tree. More...
def	upperTree (self, uTree)
	Sets upper tree.

Public Attributes
	rootNodeID
	VV is OK UU is wrong!
	treeCopulaParams
Public Attributes inherited from starvine.vine.tree.Vtree
	data
	nT
	level
	tree

Private Member Functions
def	_optimNodePairs (self)
	Selects the node-pairings which maximizes the sum over all edge weights provided the constraint of a C-tree stucture. More...
def	_evalH (self)
	Computes data set at each node for use in the next level tree. More...
def	_getEdgeCopulaParams (self, u, v)
	Get copula paramters of particular edge in tree. More...
def	_initTreeParamMap (self)
	Pack all copula paramters in the tree into a 1d numpy array for simulatneous MLE optimization. More...

Detailed Description

A C-tree is a tree with a single root node.

Each level of a cononical vine is a C-tree.

Constructor & Destructor Documentation

__init__()

def starvine.vine.C_vine.Ctree.__init__	(		self,
			data,
			lvl = None,
			kwargs
	)

A single tree within vine.

Parameters

data	DataFrame multivariate data set. Each data column will be assigned to a node.
lvl	int: tree level in the vine
weights	DataFrame: (optional) data weights
labels	list of str or ints: (optional) data labels

Member Function Documentation

_evalH()

def starvine.vine.C_vine.Ctree._evalH (   self )

private

Computes data set at each node for use in the next level tree.

Returns

DataFame : conditional distribution at tree edges.

_getEdgeCopulaParams()

def starvine.vine.C_vine.Ctree._getEdgeCopulaParams (   self,   u,   v  )

private

Get copula paramters of particular edge in tree.

Returns

np_1darray Copula parameters of edge

_initTreeParamMap()

def starvine.vine.C_vine.Ctree._initTreeParamMap (   self )

private

Pack all copula paramters in the tree into a 1d numpy array for simulatneous MLE optimization.

Sets the tree copula paramters.

_optimNodePairs()

def starvine.vine.C_vine.Ctree._optimNodePairs (   self )

private

Selects the node-pairings which maximizes the sum over all edge weights provided the constraint of a C-tree stucture.

It is feasible to try all C-tree configuations since if we have nT variables in the top level tree the number of unique C-trees is == nT.

Returns

nd_array: (nT-1, 3) shape array with PCC pairs Each row is a len 3 tuple: (rootNodeID, nodeID, kTau)

evalH()

def starvine.vine.C_vine.Ctree.evalH

(

self

)

Define nodes of the T+1 level tree.

Use the conditional distribution (h()) to obtain marginal distributions at the next tree level.

seqCopulaFit()

def starvine.vine.C_vine.Ctree.seqCopulaFit

(

self

)

Iterate through all edges in tree, fit copula models at each edge.

This is a sequential fitting operation.

See simultaneousCopulaFit() for a tree-wide simulltaneous parameter estimation.

treeNLLH()

def starvine.vine.C_vine.Ctree.treeNLLH	(		self,
			treeCopulaParams = None
	)

Compute this tree's negative log likelyhood.

For C-trees this is just the sum of copula-log-likeyhoods over all node-pairs.

Parameters

treeCopulaParams	np_1darray Copula parameter array. Contains parameters for all PCC in the tree.
paramMap	Maps edges to parameter len and location in treeCopulaParams {[u, v]: (start, Params_len_0), [u, v]: (start, Params_len), ...}

---

The documentation for this class was generated from the following file:

• starvine/vine/C_vine.py