learning.MB_MMPC_Lner.MB_MMPC_Lner Class Reference

Inheritance diagram for learning.MB_MMPC_Lner.MB_MMPC_Lner:

Collaboration diagram for learning.MB_MMPC_Lner.MB_MMPC_Lner:

Public Member Functions
def	__init__ (self, states_df, alpha, verbose=False, vtx_to_states=None, learn_later=False)
def	find_nbors (self)
def	find_PC (self, vtx=None)
def	find_MB (self, vtx=None)
Public Member Functions inherited from learning.MB_BasedLner.MB_BasedLner
def	__init__ (self, states_df, alpha, verbose=False, vtx_to_states=None, learn_later=False)
def	learn_struc (self)
def	find_MB (self, vtx=None)
def	find_nbors (self)
def	orient_edges (self)
def	new_filled_nx_graph (self)
def	undo_cycles (self)
def	orient_undecided_edges (self)
Public Member Functions inherited from learning.NetStrucLner.NetStrucLner
def	__init__ (self, is_quantum, states_df, vtx_to_states=None)
def	fill_bnet_with_parents (self, vtx_to_parents)

Public Attributes
	vtx_to_nbors
	vtx_to_MB
Public Attributes inherited from learning.MB_BasedLner.MB_BasedLner
	alpha
	verbose
	vtx_to_MB
	vtx_to_parents
	vtx_to_nbors
Public Attributes inherited from learning.NetStrucLner.NetStrucLner
	is_quantum
	bnet
	states_df
	ord_nodes

Additional Inherited Members
Static Public Member Functions inherited from learning.MB_BasedLner.MB_BasedLner
def	MB_lner_test (LnerClass, verbose=False)
Static Public Member Functions inherited from learning.NetStrucLner.NetStrucLner
def	learn_nd_state_names (bnet, states_df)
def	import_nd_state_names (bnet, vtx_to_states)
def	int_sts_detector (sub_states_df)

Detailed Description

The MMPCLner (Min Max Parents Children Learner) is a subclass of
MB_BasedLner. See docstring for MB_BasedLner for more info about this
type of algo.

Whereas the first generation of MB_ algos found the MB of each node
first and the neighbors of each node second, the MMPC algorithm finds
these in the opposite order.

PC stands for Parents-Children. The PC set of a node is the same thing as
its set of neighbors.

See Shunkai Fu Thesis for pseudo code on which this class is based.


Attributes
----------
is_quantum : bool
    True for quantum bnets amd False for classical bnets
bnet : BayesNet
    a BayesNet in which we store what is learned
states_df : pandas.DataFrame
    a Pandas DataFrame with training data. column = node and row =
    sample. Each row/sample gives the state of the col/node.
ord_nodes : list[DirectedNode]
    a list of DirectedNode's named and in the same order as the column
    labels of self.states_df.

alpha : float
    threshold used for deciding whether a conditional or unconditional
    mutual info is said to be close to zero (independence) or not (
    dependence). The error in a data entropy is on the order of ln(n+1)
    - ln(n) \approx 1/n where n is the number of samples so 5/n is a
    good default value for alpha.
verbose : bool
    True for this prints a running commentary to console
vtx_to_MB : dict[str, list[str]]
    A dictionary mapping each vertex to a list of the vertices in its
    Markov Blanket. (The MB of a node consists of its parents, children
    and children's parents, aka spouses).
vtx_to_nbors : dict[str, list[str]]
    a dictionary mapping each vertex to a list of its neighbors. The
    literature also calls the set of neighbors of a vertex its PC (
    parents-children) set.
vtx_to_parents : dict[str, list[str]]
    dictionary mapping each vertex to a list of its parents's names

Constructor & Destructor Documentation

def learning.MB_MMPC_Lner.MB_MMPC_Lner.__init__	(		self,
			states_df,
			alpha,
			verbose = False,
			vtx_to_states = None,
			learn_later = False
	)

Constructor

Parameters
----------
states_df : pandas.DatFrame
alpha : float
verbose : bool
vtx_to_states : dict[str, list[str]]
    A dictionary mapping each node name to a list of its state names.
    This information will be stored in self.bnet. If
    vtx_to_states=None, constructor will learn vtx_to_states
    from states_df
learn_later : bool
    False if you want to call the function learn_struc() inside the
    constructor. True if not.

Returns
-------
None

Member Function Documentation

def learning.MB_MMPC_Lner.MB_MMPC_Lner.find_MB	(		self,
			vtx = None
	)

This function finds the MB of vtx and stores it inside vtx_to_MB[
vtx]. If vtx=None, then it will find the MB of all the vertices of
the graph.

Parameters
----------
vtx : str

Returns
-------
None

def learning.MB_MMPC_Lner.MB_MMPC_Lner.find_nbors

(

self

)

This function is called in the parent class MB_BasedLner after the
function find_MB() is. But in this subclass we want to find the
neighbors before finding the MBs. We solve this quandary by
overriding find_nbors() so that it does nothing, and defining a new
function find_PC() that also finds the neighbors, but is called
inside find_MB(). In MB_BasedLner, the function find_nbors() assumes
that the MB of all nodes is known whereas in this subclass,
the function find_PC() doesn't assume this.

Returns
-------
None

def learning.MB_MMPC_Lner.MB_MMPC_Lner.find_PC	(		self,
			vtx = None
	)

This function finds the PC set (Parents-Children, aka neighbors) of
vtx and stores it in vtx_to_nbors. if vtx=None, it finds the PC set
of all nodes.

Parameters
----------
vtx : str

Returns
-------
None

---

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

• learning/MB_MMPC_Lner.py