Luca Moretti
4 years ago
parent
c53a65ebd7
commit
a3e614bcb8
@ -1,143 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
import numpy as np |
||||
|
||||
from ..structure_graph.network_graph import NetworkGraph |
||||
from ..structure_graph.sample_path import SetOfCims |
||||
from ..structure_graph.trajectory import Trajectory |
||||
|
||||
|
||||
class ParametersEstimator(object): |
||||
"""Has the task of computing the cims of particular node given the trajectories and the net structure |
||||
in the graph ``_net_graph``. |
||||
|
||||
:param trajectories: the trajectories |
||||
:type trajectories: Trajectory |
||||
:param net_graph: the net structure |
||||
:type net_graph: NetworkGraph |
||||
:_single_set_of_cims: the set of cims object that will hold the cims of the node |
||||
""" |
||||
|
||||
def __init__(self, trajectories: Trajectory, net_graph: NetworkGraph): |
||||
"""Constructor Method |
||||
""" |
||||
self._trajectories = trajectories |
||||
self._net_graph = net_graph |
||||
self._single_set_of_cims = None |
||||
|
||||
def fast_init(self, node_id: str) -> None: |
||||
"""Initializes all the necessary structures for the parameters estimation for the node ``node_id``. |
||||
|
||||
:param node_id: the node label |
||||
:type node_id: string |
||||
""" |
||||
p_vals = self._net_graph._aggregated_info_about_nodes_parents[2] |
||||
node_states_number = self._net_graph.get_states_number(node_id) |
||||
self._single_set_of_cims = SetOfCims(node_id, p_vals, node_states_number, self._net_graph.p_combs) |
||||
|
||||
def compute_parameters_for_node(self, node_id: str) -> SetOfCims: |
||||
"""Compute the CIMS of the node identified by the label ``node_id``. |
||||
|
||||
:param node_id: the node label |
||||
:type node_id: string |
||||
:return: A SetOfCims object filled with the computed CIMS |
||||
:rtype: SetOfCims |
||||
""" |
||||
node_indx = self._net_graph.get_node_indx(node_id) |
||||
state_res_times = self._single_set_of_cims._state_residence_times |
||||
transition_matrices = self._single_set_of_cims._transition_matrices |
||||
ParametersEstimator.compute_state_res_time_for_node(self._trajectories.times, |
||||
self._trajectories.trajectory, |
||||
self._net_graph.time_filtering, |
||||
self._net_graph.time_scalar_indexing_strucure, |
||||
state_res_times) |
||||
ParametersEstimator.compute_state_transitions_for_a_node(node_indx, self._trajectories.complete_trajectory, |
||||
self._net_graph.transition_filtering, |
||||
self._net_graph.transition_scalar_indexing_structure, |
||||
transition_matrices) |
||||
self._single_set_of_cims.build_cims(state_res_times, transition_matrices) |
||||
return self._single_set_of_cims |
||||
|
||||
@staticmethod |
||||
def compute_state_res_time_for_node(times: np.ndarray, trajectory: np.ndarray, |
||||
cols_filter: np.ndarray, scalar_indexes_struct: np.ndarray, |
||||
T: np.ndarray) -> None: |
||||
"""Compute the state residence times for a node and fill the matrix ``T`` with the results |
||||
|
||||
:param node_indx: the index of the node |
||||
:type node_indx: int |
||||
:param times: the times deltas vector |
||||
:type times: numpy.array |
||||
:param trajectory: the trajectory |
||||
:type trajectory: numpy.ndArray |
||||
:param cols_filter: the columns filtering structure |
||||
:type cols_filter: numpy.array |
||||
:param scalar_indexes_struct: the indexing structure |
||||
:type scalar_indexes_struct: numpy.array |
||||
:param T: the state residence times vectors |
||||
:type T: numpy.ndArray |
||||
""" |
||||
T[:] = np.bincount(np.sum(trajectory[:, cols_filter] * scalar_indexes_struct / scalar_indexes_struct[0], axis=1) |
||||
.astype(np.int), \ |
||||
times, |
||||
minlength=scalar_indexes_struct[-1]).reshape(-1, T.shape[1]) |
||||
|
||||
@staticmethod |
||||
def compute_state_transitions_for_a_node(node_indx: int, trajectory: np.ndarray, cols_filter: np.ndarray, |
||||
scalar_indexing: np.ndarray, M: np.ndarray) -> None: |
||||
"""Compute the state residence times for a node and fill the matrices ``M`` with the results. |
||||
|
||||
:param node_indx: the index of the node |
||||
:type node_indx: int |
||||
:param trajectory: the trajectory |
||||
:type trajectory: numpy.ndArray |
||||
:param cols_filter: the columns filtering structure |
||||
:type cols_filter: numpy.array |
||||
:param scalar_indexing: the indexing structure |
||||
:type scalar_indexing: numpy.array |
||||
:param M: the state transitions matrices |
||||
:type M: numpy.ndArray |
||||
""" |
||||
diag_indices = np.array([x * M.shape[1] + x % M.shape[1] for x in range(M.shape[0] * M.shape[1])], |
||||
dtype=np.int64) |
||||
trj_tmp = trajectory[trajectory[:, int(trajectory.shape[1] / 2) + node_indx].astype(np.int) >= 0] |
||||
M[:] = np.bincount(np.sum(trj_tmp[:, cols_filter] * scalar_indexing / scalar_indexing[0], axis=1).astype(np.int) |
||||
, minlength=scalar_indexing[-1]).reshape(-1, M.shape[1], M.shape[2]) |
||||
M_raveled = M.ravel() |
||||
M_raveled[diag_indices] = 0 |
||||
M_raveled[diag_indices] = np.sum(M, axis=2).ravel() |
||||
|
||||
def init_sets_cims_container(self): |
||||
self.sets_of_cims_struct = acims.SetsOfCimsContainer(self.net_graph.nodes, |
||||
self.net_graph.nodes_values, |
||||
self.net_graph.get_ordered_by_indx_parents_values_for_all_nodes(), |
||||
self.net_graph.p_combs) |
||||
|
||||
def compute_parameters(self): |
||||
#print(self.net_graph.get_nodes()) |
||||
#print(self.amalgamated_cims_struct.sets_of_cims) |
||||
#enumerate(zip(self.net_graph.get_nodes(), self.amalgamated_cims_struct.sets_of_cims)) |
||||
for indx, aggr in enumerate(zip(self.net_graph.nodes, self.sets_of_cims_struct.sets_of_cims)): |
||||
#print(self.net_graph.time_filtering[indx]) |
||||
#print(self.net_graph.time_scalar_indexing_strucure[indx]) |
||||
self.compute_state_res_time_for_node(self.net_graph.get_node_indx(aggr[0]), self.sample_path.trajectories.times, |
||||
self.sample_path.trajectories.trajectory, |
||||
self.net_graph.time_filtering[indx], |
||||
self.net_graph.time_scalar_indexing_strucure[indx], |
||||
aggr[1]._state_residence_times) |
||||
#print(self.net_graph.transition_filtering[indx]) |
||||
#print(self.net_graph.transition_scalar_indexing_structure[indx]) |
||||
self.compute_state_transitions_for_a_node(self.net_graph.get_node_indx(aggr[0]), |
||||
self.sample_path.trajectories.complete_trajectory, |
||||
self.net_graph.transition_filtering[indx], |
||||
self.net_graph.transition_scalar_indexing_structure[indx], |
||||
aggr[1]._transition_matrices) |
||||
aggr[1].build_cims(aggr[1]._state_residence_times, aggr[1]._transition_matrices) |
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
@ -1,245 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
import itertools |
||||
import json |
||||
import typing |
||||
|
||||
import networkx as nx |
||||
import numpy as np |
||||
from networkx.readwrite import json_graph |
||||
import os |
||||
from scipy.stats import chi2 as chi2_dist |
||||
from scipy.stats import f as f_dist |
||||
from tqdm import tqdm |
||||
|
||||
from ..utility.cache as ch |
||||
from ..structure_graph.conditional_intensity_matrix import ConditionalIntensityMatrix |
||||
from ..structure_graph.network_graph import NetworkGraph |
||||
from .parameters_estimator import ParametersEstimator |
||||
from .structure_estimator import StructureEstimator |
||||
from ..structure_graph.sample_path import SamplePath |
||||
from ..structure_graph.structure import Structure |
||||
from ..optimizers.constraint_based_optimizer import ConstraintBasedOptimizer |
||||
|
||||
import concurrent.futures |
||||
|
||||
from utility.decorators import timing,timing_write |
||||
|
||||
import multiprocessing |
||||
from multiprocessing import Pool |
||||
|
||||
|
||||
class StructureConstraintBasedEstimator(se.StructureEstimator): |
||||
""" |
||||
Has the task of estimating the network structure given the trajectories in samplepath by using a constraint-based approach. |
||||
|
||||
:param sample_path: the _sample_path object containing the trajectories and the real structure |
||||
:type sample_path: SamplePath |
||||
:param exp_test_alfa: the significance level for the exponential Hp test |
||||
:type exp_test_alfa: float |
||||
:param chi_test_alfa: the significance level for the chi Hp test |
||||
:type chi_test_alfa: float |
||||
:_nodes: the nodes labels |
||||
:_nodes_vals: the nodes cardinalities |
||||
:_nodes_indxs: the nodes indexes |
||||
:_complete_graph: the complete directed graph built using the nodes labels in ``_nodes`` |
||||
:_cache: the Cache object |
||||
""" |
||||
|
||||
def __init__(self, sample_path: SamplePath, exp_test_alfa: float, chi_test_alfa: float,known_edges: typing.List= [],thumb_threshold:int = 25): |
||||
super().__init__(sample_path,known_edges) |
||||
self._exp_test_sign = exp_test_alfa |
||||
self._chi_test_alfa = chi_test_alfa |
||||
self._thumb_threshold = thumb_threshold |
||||
tot_vars_count: int, parent_indx, child_indx) -> bool: |
||||
|
||||
def complete_test(self, test_parent: str, test_child: str, parent_set: typing.List, child_states_numb: int, |
||||
tot_vars_count: int, parent_indx, child_indx) -> bool: |
||||
"""Performs a complete independence test on the directed graphs G1 = {test_child U parent_set} |
||||
G2 = {G1 U test_parent} (added as an additional parent of the test_child). |
||||
Generates all the necessary structures and datas to perform the tests. |
||||
|
||||
:param test_parent: the node label of the test parent |
||||
:type test_parent: string |
||||
:param test_child: the node label of the child |
||||
:type test_child: string |
||||
:param parent_set: the common parent set |
||||
:type parent_set: List |
||||
:param child_states_numb: the cardinality of the ``test_child`` |
||||
:type child_states_numb: int |
||||
:param tot_vars_count: the total number of variables in the net |
||||
:type tot_vars_count: int |
||||
:return: True iff test_child and test_parent are independent given the sep_set parent_set. False otherwise |
||||
:rtype: bool |
||||
""" |
||||
p_set = parent_set[:] |
||||
complete_info = parent_set[:] |
||||
complete_info.append(test_child) |
||||
|
||||
parents = np.array(parent_set) |
||||
parents = np.append(parents, test_parent) |
||||
sorted_parents = self._nodes[np.isin(self._nodes, parents)] |
||||
cims_filter = sorted_parents != test_parent |
||||
|
||||
p_set.insert(0, test_parent) |
||||
sofc2 = self._cache.find(set(p_set)) |
||||
|
||||
if not sofc2: |
||||
complete_info.append(test_parent) |
||||
bool_mask2 = np.isin(self._nodes, complete_info) |
||||
l2 = list(self._nodes[bool_mask2]) |
||||
indxs2 = self._nodes_indxs[bool_mask2] |
||||
vals2 = self._nodes_vals[bool_mask2] |
||||
eds2 = list(itertools.product(p_set, test_child)) |
||||
s2 = Structure(l2, indxs2, vals2, eds2, tot_vars_count) |
||||
g2 = NetworkGraph(s2) |
||||
g2.fast_init(test_child) |
||||
p2 = ParametersEstimator(self._sample_path.trajectories, g2) |
||||
p2.fast_init(test_child) |
||||
sofc2 = p2.compute_parameters_for_node(test_child) |
||||
self._cache.put(set(p_set), sofc2) |
||||
|
||||
del p_set[0] |
||||
sofc1 = self._cache.find(set(p_set)) |
||||
if not sofc1: |
||||
g2.remove_node(test_parent) |
||||
g2.fast_init(test_child) |
||||
p2 = ParametersEstimator(self._sample_path.trajectories, g2) |
||||
p2.fast_init(test_child) |
||||
sofc1 = p2.compute_parameters_for_node(test_child) |
||||
self._cache.put(set(p_set), sofc1) |
||||
thumb_value = 0.0 |
||||
if child_states_numb > 2: |
||||
parent_val = self._sample_path.structure.get_states_number(test_parent) |
||||
bool_mask_vals = np.isin(self._nodes, parent_set) |
||||
parents_vals = self._nodes_vals[bool_mask_vals] |
||||
thumb_value = self.compute_thumb_value(parent_val, child_states_numb, parents_vals) |
||||
for cim1, p_comb in zip(sofc1.actual_cims, sofc1.p_combs): |
||||
cond_cims = sofc2.filter_cims_with_mask(cims_filter, p_comb) |
||||
for cim2 in cond_cims: |
||||
if not self.independence_test(child_states_numb, cim1, cim2, thumb_value, parent_indx, child_indx): |
||||
return False |
||||
return True |
||||
|
||||
def independence_test(self, child_states_numb: int, cim1: ConditionalIntensityMatrix, |
||||
cim2: ConditionalIntensityMatrix, thumb_value: float, parent_indx, child_indx) -> bool: |
||||
"""Compute the actual independence test using two cims. |
||||
It is performed first the exponential test and if the null hypothesis is not rejected, |
||||
it is performed also the chi_test. |
||||
|
||||
:param child_states_numb: the cardinality of the test child |
||||
:type child_states_numb: int |
||||
:param cim1: a cim belonging to the graph without test parent |
||||
:type cim1: ConditionalIntensityMatrix |
||||
:param cim2: a cim belonging to the graph with test parent |
||||
:type cim2: ConditionalIntensityMatrix |
||||
:return: True iff both tests do NOT reject the null hypothesis of independence. False otherwise. |
||||
:rtype: bool |
||||
""" |
||||
M1 = cim1.state_transition_matrix |
||||
M2 = cim2.state_transition_matrix |
||||
r1s = M1.diagonal() |
||||
r2s = M2.diagonal() |
||||
C1 = cim1.cim |
||||
C2 = cim2.cim |
||||
if child_states_numb > 2: |
||||
if (np.sum(np.diagonal(M1)) / thumb_value) < self._thumb_threshold: |
||||
self._removable_edges_matrix[parent_indx][child_indx] = False |
||||
return False |
||||
F_stats = C2.diagonal() / C1.diagonal() |
||||
exp_alfa = self._exp_test_sign |
||||
for val in range(0, child_states_numb): |
||||
if F_stats[val] < f_dist.ppf(exp_alfa / 2, r1s[val], r2s[val]) or \ |
||||
F_stats[val] > f_dist.ppf(1 - exp_alfa / 2, r1s[val], r2s[val]): |
||||
return False |
||||
M1_no_diag = M1[~np.eye(M1.shape[0], dtype=bool)].reshape(M1.shape[0], -1) |
||||
M2_no_diag = M2[~np.eye(M2.shape[0], dtype=bool)].reshape( |
||||
M2.shape[0], -1) |
||||
chi_2_quantile = chi2_dist.ppf(1 - self._chi_test_alfa, child_states_numb - 1) |
||||
Ks = np.sqrt(r1s / r2s) |
||||
Ls = np.sqrt(r2s / r1s) |
||||
for val in range(0, child_states_numb): |
||||
Chi = np.sum(np.power(Ks[val] * M2_no_diag[val] - Ls[val] *M1_no_diag[val], 2) / |
||||
(M1_no_diag[val] + M2_no_diag[val])) |
||||
if Chi > chi_2_quantile: |
||||
return False |
||||
return True |
||||
|
||||
def compute_thumb_value(self, parent_val, child_val, parent_set_vals): |
||||
"""Compute the value to test against the thumb_threshold. |
||||
|
||||
:param parent_val: test parent's variable cardinality |
||||
:type parent_val: int |
||||
:param child_val: test child's variable cardinality |
||||
:type child_val: int |
||||
:param parent_set_vals: the cardinalities of the nodes in the current sep-set |
||||
:type parent_set_vals: List |
||||
:return: the thumb value for the current independence test |
||||
:rtype: int |
||||
""" |
||||
df = (child_val - 1) ** 2 |
||||
df = df * parent_val |
||||
for v in parent_set_vals: |
||||
df = df * v |
||||
return df |
||||
|
||||
def one_iteration_of_CTPC_algorithm(self, var_id: str, tot_vars_count: int)-> typing.List: |
||||
"""Performs an iteration of the CTPC algorithm using the node ``var_id`` as ``test_child``. |
||||
|
||||
:param var_id: the node label of the test child |
||||
:type var_id: string |
||||
""" |
||||
optimizer_obj = optimizer.ConstraintBasedOptimizer( |
||||
node_id = var_id, |
||||
structure_estimator = self, |
||||
tot_vars_count = tot_vars_count) |
||||
return optimizer_obj.optimize_structure() |
||||
|
||||
|
||||
def ctpc_algorithm(self,disable_multiprocessing:bool= False ): |
||||
""" |
||||
Compute the CTPC algorithm. |
||||
Parameters: |
||||
void |
||||
Returns: |
||||
void |
||||
""" |
||||
ctpc_algo = self.one_iteration_of_CTPC_algorithm |
||||
total_vars_numb = self._sample_path.total_variables_count |
||||
|
||||
n_nodes= len(self.nodes) |
||||
|
||||
total_vars_numb_array = [total_vars_numb] * n_nodes |
||||
|
||||
'get the number of CPU' |
||||
cpu_count = multiprocessing.cpu_count() |
||||
|
||||
|
||||
|
||||
'Remove all the edges from the structure' |
||||
self._sample_path.structure.clean_structure_edges() |
||||
|
||||
'Estimate the best parents for each node' |
||||
#with multiprocessing.Pool(processes=cpu_count) as pool: |
||||
#with get_context("spawn").Pool(processes=cpu_count) as pool: |
||||
if disable_multiprocessing: |
||||
print("DISABILITATO") |
||||
cpu_count = 1 |
||||
list_edges_partial = [ctpc_algo(n,total_vars_numb) for n in self.nodes] |
||||
else: |
||||
with concurrent.futures.ProcessPoolExecutor(max_workers=cpu_count) as executor: |
||||
list_edges_partial = executor.map(ctpc_algo, |
||||
self.nodes, |
||||
total_vars_numb_array) |
||||
#list_edges_partial = [ctpc_algo(n,total_vars_numb) for n in self.nodes] |
||||
|
||||
return set(itertools.chain.from_iterable(list_edges_partial)) |
||||
|
||||
|
||||
@timing |
||||
def estimate_structure(self,disable_multiprocessing:bool=False): |
||||
return self.ctpc_algorithm(disable_multiprocessing=disable_multiprocessing) |
||||
|
||||
|
||||
|
||||
|
||||
@ -1,189 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
import itertools |
||||
import json |
||||
import typing |
||||
|
||||
import matplotlib.pyplot as plt |
||||
import networkx as nx |
||||
import numpy as np |
||||
from networkx.readwrite import json_graph |
||||
|
||||
from abc import ABC |
||||
|
||||
import abc |
||||
|
||||
import ..utility.cache as ch |
||||
import ..structure_graph.conditional_intensity_matrix import ConditionalIntensityMatrix |
||||
import ..structure_graph.network_graph import NetworkGraph |
||||
import .parameters_estimator import ParametersEstimator |
||||
import ..structure_graph.sample_path import SamplePath |
||||
from ..structure_graph.structure import Structure |
||||
|
||||
|
||||
class StructureEstimator(object): |
||||
"""Has the task of estimating the network structure given the trajectories in ``samplepath``. |
||||
|
||||
:param sample_path: the _sample_path object containing the trajectories and the real structure |
||||
:type sample_path: SamplePath |
||||
:_nodes: the nodes labels |
||||
:_nodes_vals: the nodes cardinalities |
||||
:_nodes_indxs: the nodes indexes |
||||
:_complete_graph: the complete directed graph built using the nodes labels in ``_nodes`` |
||||
""" |
||||
|
||||
def __init__(self, sample_path: SamplePath, known_edges: typing.List = None): |
||||
self._sample_path = sample_path |
||||
self.nodes = np.array(self._sample_path.structure.nodes_labels) |
||||
self.nodes_vals = self._sample_path.structure.nodes_values |
||||
self.nodes_indxs = self._sample_path.structure.nodes_indexes |
||||
self._removable_edges_matrix = self. |
||||
(known_edges) |
||||
self.complete_graph = self.build_complete_graph(self._sample_path.structure.nodes_labels) |
||||
self.cache = ch.Cache() |
||||
|
||||
def build_removable_edges_matrix(self, known_edges: typing.List): |
||||
"""Builds a boolean matrix who shows if a edge could be removed or not, based on prior knowledge given: |
||||
|
||||
:param known_edges: the list of nodes labels |
||||
:type known_edges: List |
||||
:return: a boolean matrix |
||||
:rtype: np.ndarray |
||||
""" |
||||
tot_vars_count = self._sample_path.total_variables_count |
||||
complete_adj_matrix = np.full((tot_vars_count, tot_vars_count), True) |
||||
if known_edges: |
||||
for edge in known_edges: |
||||
i = self._sample_path.structure.get_node_indx(edge[0]) |
||||
j = self._sample_path.structure.get_node_indx(edge[1]) |
||||
complete_adj_matrix[i][j] = False |
||||
return complete_adj_matrix |
||||
|
||||
@staticmethod |
||||
def build_complete_graph(node_ids: typing.List) -> nx.DiGraph: |
||||
"""Builds a complete directed graph (no self loops) given the nodes labels in the list ``node_ids``: |
||||
|
||||
:param node_ids: the list of nodes labels |
||||
:type node_ids: List |
||||
:return: a complete Digraph Object |
||||
:rtype: networkx.DiGraph |
||||
""" |
||||
complete_graph = nx.DiGraph() |
||||
complete_graph.add_nodes_from(node_ids) |
||||
complete_graph.add_edges_from(itertools.permutations(node_ids, 2)) |
||||
return complete_graph |
||||
|
||||
|
||||
|
||||
def generate_possible_sub_sets_of_size(self, u: typing.List, size: int, parent_label: str): |
||||
"""Creates a list containing all possible subsets of the list ``u`` of size ``size``, |
||||
that do not contains a the node identified by ``parent_label``. |
||||
|
||||
:param u: the list of nodes |
||||
:type u: List |
||||
:param size: the size of the subsets |
||||
:type size: int |
||||
:param parent_label: the node to exclude in the subsets generation |
||||
:type parent_label: string |
||||
:return: an Iterator Object containing a list of lists |
||||
:rtype: Iterator |
||||
""" |
||||
list_without_test_parent = u[:] |
||||
list_without_test_parent.remove(parent_label) |
||||
return map(list, itertools.combinations(list_without_test_parent, size)) |
||||
|
||||
def save_results(self) -> None: |
||||
"""Save the estimated Structure to a .json file in the path where the data are loaded from. |
||||
The file is named as the input dataset but the `results_` word is appended to the results file. |
||||
""" |
||||
res = json_graph.node_link_data(self._complete_graph) |
||||
name = self._sample_path._importer.file_path.rsplit('/', 1)[-1] |
||||
name = name.split('.', 1)[0] |
||||
name += '_' + str(self._sample_path._importer.dataset_id()) |
||||
name += '.json' |
||||
file_name = 'results_' + name |
||||
with open(file_name, 'w') as f: |
||||
json.dump(res, f) |
||||
|
||||
|
||||
def remove_diagonal_elements(self, matrix): |
||||
m = matrix.shape[0] |
||||
strided = np.lib.stride_tricks.as_strided |
||||
s0, s1 = matrix.strides |
||||
return strided(matrix.ravel()[1:], shape=(m - 1, m), strides=(s0 + s1, s1)).reshape(m, -1) |
||||
|
||||
|
||||
@abc.abstractmethod |
||||
def estimate_structure(self) -> typing.List: |
||||
"""Abstract method to estimate the structure |
||||
|
||||
:return: List of estimated edges |
||||
:rtype: Typing.List |
||||
""" |
||||
pass |
||||
|
||||
|
||||
def adjacency_matrix(self) -> np.ndarray: |
||||
"""Converts the estimated structure ``_complete_graph`` to a boolean adjacency matrix representation. |
||||
|
||||
:return: The adjacency matrix of the graph ``_complete_graph`` |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
return nx.adj_matrix(self._complete_graph).toarray().astype(bool) |
||||
|
||||
def spurious_edges(self) -> typing.List: |
||||
"""Return the spurious edges present in the estimated structure, if a prior net structure is present in |
||||
``_sample_path.structure``. |
||||
|
||||
:return: A list containing the spurious edges |
||||
:rtype: List |
||||
""" |
||||
if not self._sample_path.has_prior_net_structure: |
||||
raise RuntimeError("Can not compute spurious edges with no prior net structure!") |
||||
real_graph = nx.DiGraph() |
||||
real_graph.add_nodes_from(self._sample_path.structure.nodes_labels) |
||||
real_graph.add_edges_from(self._sample_path.structure.edges) |
||||
return nx.difference(real_graph, self._complete_graph).edges |
||||
|
||||
def save_plot_estimated_structure_graph(self) -> None: |
||||
"""Plot the estimated structure in a graphical model style. |
||||
Spurious edges are colored in red. |
||||
""" |
||||
graph_to_draw = nx.DiGraph() |
||||
spurious_edges = self.spurious_edges() |
||||
non_spurious_edges = list(set(self._complete_graph.edges) - set(spurious_edges)) |
||||
print(non_spurious_edges) |
||||
edges_colors = ['red' if edge in spurious_edges else 'black' for edge in self._complete_graph.edges] |
||||
graph_to_draw.add_edges_from(spurious_edges) |
||||
graph_to_draw.add_edges_from(non_spurious_edges) |
||||
pos = nx.spring_layout(graph_to_draw, k=0.5*1/np.sqrt(len(graph_to_draw.nodes())), iterations=50,scale=10) |
||||
options = { |
||||
"node_size": 2000, |
||||
"node_color": "white", |
||||
"edgecolors": "black", |
||||
'linewidths':2, |
||||
"with_labels":True, |
||||
"font_size":13, |
||||
'connectionstyle': 'arc3, rad = 0.1', |
||||
"arrowsize": 15, |
||||
"arrowstyle": '<|-', |
||||
"width": 1, |
||||
"edge_color":edges_colors, |
||||
} |
||||
|
||||
nx.draw(graph_to_draw, pos, **options) |
||||
ax = plt.gca() |
||||
ax.margins(0.20) |
||||
plt.axis("off") |
||||
name = self._sample_path._importer.file_path.rsplit('/', 1)[-1] |
||||
name = name.split('.', 1)[0] |
||||
name += '_' + str(self._sample_path._importer.dataset_id()) |
||||
name += '.png' |
||||
plt.savefig(name) |
||||
plt.clf() |
||||
print("Estimated Structure Plot Saved At: ", os.path.abspath(name)) |
||||
|
||||
|
||||
|
||||
|
||||
|
||||
@ -1,41 +0,0 @@ |
||||
from abc import ABC, abstractmethod |
||||
|
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
import utility.abstract_importer as ai |
||||
|
||||
|
||||
class AbstractSamplePath(ABC): |
||||
|
||||
def __init__(self, importer: ai.AbstractImporter): |
||||
self.importer = importer |
||||
self._trajectories = None |
||||
self._structure = None |
||||
super().__init__() |
||||
|
||||
@abstractmethod |
||||
def build_trajectories(self): |
||||
""" |
||||
Builds the Trajectory object that will contain all the trajectories. |
||||
Assigns the Trajectoriy object to the instance attribute _trajectories |
||||
Clears all the unused dataframes in Importer Object |
||||
|
||||
Parameters: |
||||
void |
||||
Returns: |
||||
void |
||||
""" |
||||
pass |
||||
|
||||
@abstractmethod |
||||
def build_structure(self): |
||||
""" |
||||
Builds the Structure object that aggregates all the infos about the net. |
||||
Assigns the Structure object to the instance attribuite _structure |
||||
Parameters: |
||||
void |
||||
Returns: |
||||
void |
||||
""" |
||||
pass |
||||
@ -1,44 +0,0 @@ |
||||
import numpy as np |
||||
|
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
class ConditionalIntensityMatrix: |
||||
"""Abstracts the Conditional Intesity matrix of a node as aggregation of the state residence times vector |
||||
and state transition matrix and the actual CIM matrix. |
||||
|
||||
:param state_residence_times: state residence times vector |
||||
:type state_residence_times: numpy.array |
||||
:param state_transition_matrix: the transitions count matrix |
||||
:type state_transition_matrix: numpy.ndArray |
||||
:_cim: the actual cim of the node |
||||
""" |
||||
def __init__(self, state_residence_times: np.array, state_transition_matrix: np.array): |
||||
"""Constructor Method |
||||
""" |
||||
self._state_residence_times = state_residence_times |
||||
self._state_transition_matrix = state_transition_matrix |
||||
self._cim = self.state_transition_matrix.astype(np.float64) |
||||
|
||||
def compute_cim_coefficients(self): |
||||
"""Compute the coefficients of the matrix _cim by using the following equality q_xx' = M[x, x'] / T[x]. |
||||
The class member ``_cim`` will contain the computed cim |
||||
""" |
||||
np.fill_diagonal(self._cim, self._cim.diagonal() * -1) |
||||
self._cim = ((self._cim.T + 1) / (self._state_residence_times + 1)).T |
||||
|
||||
@property |
||||
def state_residence_times(self): |
||||
return self._state_residence_times |
||||
|
||||
@property |
||||
def state_transition_matrix(self): |
||||
return self._state_transition_matrix |
||||
|
||||
@property |
||||
def cim(self): |
||||
return self._cim |
||||
|
||||
def __repr__(self): |
||||
return 'CIM:\n' + str(self.cim) |
||||
|
||||
@ -1,285 +0,0 @@ |
||||
|
||||
import typing |
||||
|
||||
import networkx as nx |
||||
import numpy as np |
||||
|
||||
from .structure import Structure |
||||
|
||||
|
||||
class NetworkGraph(object): |
||||
"""Abstracts the infos contained in the Structure class in the form of a directed graph. |
||||
Has the task of creating all the necessary filtering and indexing structures for parameters estimation |
||||
|
||||
:param graph_struct: the ``Structure`` object from which infos about the net will be extracted |
||||
:type graph_struct: Structure |
||||
:_graph: directed graph |
||||
:_aggregated_info_about_nodes_parents: a structure that contains all the necessary infos |
||||
about every parents of the node of which all the indexing and filtering structures will be constructed. |
||||
:_time_scalar_indexing_structure: the indexing structure for state res time estimation |
||||
:_transition_scalar_indexing_structure: the indexing structure for transition computation |
||||
:_time_filtering: the columns filtering structure used in the computation of the state res times |
||||
:_transition_filtering: the columns filtering structure used in the computation of the transition |
||||
from one state to another |
||||
:_p_combs_structure: all the possible parents states combination for the node of interest |
||||
""" |
||||
|
||||
def __init__(self, graph_struct: Structure): |
||||
"""Constructor Method |
||||
""" |
||||
self._graph_struct = graph_struct |
||||
self._graph = nx.DiGraph() |
||||
self._aggregated_info_about_nodes_parents = None |
||||
self._time_scalar_indexing_structure = None |
||||
self._transition_scalar_indexing_structure = None |
||||
self._time_filtering = None |
||||
self._transition_filtering = None |
||||
self._p_combs_structure = None |
||||
|
||||
def init_graph(self): |
||||
self.add_nodes(self._nodes_labels) |
||||
self.add_edges(self.graph_struct.edges) |
||||
self.aggregated_info_about_nodes_parents = self.get_ord_set_of_par_of_all_nodes() |
||||
self._fancy_indexing = self.build_fancy_indexing_structure(0) |
||||
self.build_scalar_indexing_structures() |
||||
self.build_time_columns_filtering_structure() |
||||
self.build_transition_columns_filtering_structure() |
||||
self._p_combs_structure = self.build_p_combs_structure() |
||||
|
||||
def fast_init(self, node_id: str) -> None: |
||||
"""Initializes all the necessary structures for parameters estimation of the node identified by the label |
||||
node_id |
||||
|
||||
:param node_id: the label of the node |
||||
:type node_id: string |
||||
""" |
||||
self.add_nodes(self._graph_struct.nodes_labels) |
||||
self.add_edges(self._graph_struct.edges) |
||||
self._aggregated_info_about_nodes_parents = self.get_ordered_by_indx_set_of_parents(node_id) |
||||
p_indxs = self._aggregated_info_about_nodes_parents[1] |
||||
p_vals = self._aggregated_info_about_nodes_parents[2] |
||||
node_states = self.get_states_number(node_id) |
||||
node_indx = self.get_node_indx(node_id) |
||||
cols_number = self._graph_struct.total_variables_number |
||||
self._time_scalar_indexing_structure = NetworkGraph.\ |
||||
build_time_scalar_indexing_structure_for_a_node(node_states, p_vals) |
||||
self._transition_scalar_indexing_structure = NetworkGraph.\ |
||||
build_transition_scalar_indexing_structure_for_a_node(node_states, p_vals) |
||||
self._time_filtering = NetworkGraph.build_time_columns_filtering_for_a_node(node_indx, p_indxs) |
||||
self._transition_filtering = NetworkGraph.build_transition_filtering_for_a_node(node_indx, p_indxs, cols_number) |
||||
self._p_combs_structure = NetworkGraph.build_p_comb_structure_for_a_node(p_vals) |
||||
|
||||
def add_nodes(self, list_of_nodes: typing.List) -> None: |
||||
"""Adds the nodes to the ``_graph`` contained in the list of nodes ``list_of_nodes``. |
||||
Sets all the properties that identify a nodes (index, positional index, cardinality) |
||||
|
||||
:param list_of_nodes: the nodes to add to ``_graph`` |
||||
:type list_of_nodes: List |
||||
""" |
||||
nodes_indxs = self._graph_struct.nodes_indexes |
||||
nodes_vals = self._graph_struct.nodes_values |
||||
pos = 0 |
||||
for id, node_indx, node_val in zip(list_of_nodes, nodes_indxs, nodes_vals): |
||||
self._graph.add_node(id, indx=node_indx, val=node_val, pos_indx=pos) |
||||
pos += 1 |
||||
|
||||
def has_edge(self,edge:tuple)-> bool: |
||||
""" |
||||
Check if the graph contains a specific edge |
||||
|
||||
Parameters: |
||||
edge: a tuple that rappresents the edge |
||||
Returns: |
||||
bool |
||||
""" |
||||
return self.graph.has_edge(edge[0],edge[1]) |
||||
|
||||
def add_edges(self, list_of_edges: typing.List) -> None: |
||||
"""Add the edges to the ``_graph`` contained in the list ``list_of_edges``. |
||||
|
||||
:param list_of_edges: the list containing of tuples containing the edges |
||||
:type list_of_edges: List |
||||
""" |
||||
self._graph.add_edges_from(list_of_edges) |
||||
|
||||
def remove_node(self, node_id: str) -> None: |
||||
"""Remove the node ``node_id`` from all the class members. |
||||
Initialize all the filtering/indexing structures. |
||||
""" |
||||
self._graph.remove_node(node_id) |
||||
self._graph_struct.remove_node(node_id) |
||||
self.clear_indexing_filtering_structures() |
||||
|
||||
def clear_indexing_filtering_structures(self) -> None: |
||||
"""Initialize all the filtering/indexing structures. |
||||
""" |
||||
self._aggregated_info_about_nodes_parents = None |
||||
self._time_scalar_indexing_structure = None |
||||
self._transition_scalar_indexing_structure = None |
||||
self._time_filtering = None |
||||
self._transition_filtering = None |
||||
self._p_combs_structure = None |
||||
|
||||
def get_ordered_by_indx_set_of_parents(self, node: str) -> typing.Tuple: |
||||
"""Builds the aggregated structure that holds all the infos relative to the parent set of the node, namely |
||||
(parents_labels, parents_indexes, parents_cardinalities). |
||||
|
||||
:param node: the label of the node |
||||
:type node: string |
||||
:return: a tuple containing all the parent set infos |
||||
:rtype: Tuple |
||||
""" |
||||
parents = self.get_parents_by_id(node) |
||||
nodes = self._graph_struct.nodes_labels |
||||
d = {v: i for i, v in enumerate(nodes)} |
||||
sorted_parents = sorted(parents, key=lambda v: d[v]) |
||||
get_node_indx = self.get_node_indx |
||||
p_indxes = [get_node_indx(node) for node in sorted_parents] |
||||
p_values = [self.get_states_number(node) for node in sorted_parents] |
||||
return sorted_parents, p_indxes, p_values |
||||
|
||||
@staticmethod |
||||
def build_time_scalar_indexing_structure_for_a_node(node_states: int, |
||||
parents_vals: typing.List) -> np.ndarray: |
||||
"""Builds an indexing structure for the computation of state residence times values. |
||||
|
||||
:param node_states: the node cardinality |
||||
:type node_states: int |
||||
:param parents_vals: the caridinalites of the node's parents |
||||
:type parents_vals: List |
||||
:return: The time indexing structure |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
T_vector = np.array([node_states]) |
||||
T_vector = np.append(T_vector, parents_vals) |
||||
T_vector = T_vector.cumprod().astype(np.int) |
||||
return T_vector |
||||
|
||||
@staticmethod |
||||
def build_transition_scalar_indexing_structure_for_a_node(node_states_number: int, parents_vals: typing.List) \ |
||||
-> np.ndarray: |
||||
"""Builds an indexing structure for the computation of state transitions values. |
||||
|
||||
:param node_states_number: the node cardinality |
||||
:type node_states_number: int |
||||
:param parents_vals: the caridinalites of the node's parents |
||||
:type parents_vals: List |
||||
:return: The transition indexing structure |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
M_vector = np.array([node_states_number, |
||||
node_states_number]) |
||||
M_vector = np.append(M_vector, parents_vals) |
||||
M_vector = M_vector.cumprod().astype(np.int) |
||||
return M_vector |
||||
|
||||
@staticmethod |
||||
def build_time_columns_filtering_for_a_node(node_indx: int, p_indxs: typing.List) -> np.ndarray: |
||||
""" |
||||
Builds the necessary structure to filter the desired columns indicated by ``node_indx`` and ``p_indxs`` |
||||
in the dataset. |
||||
This structute will be used in the computation of the state res times. |
||||
:param node_indx: the index of the node |
||||
:type node_indx: int |
||||
:param p_indxs: the indexes of the node's parents |
||||
:type p_indxs: List |
||||
:return: The filtering structure for times estimation |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
return np.append(np.array([node_indx], dtype=np.int), p_indxs).astype(np.int) |
||||
|
||||
@staticmethod |
||||
def build_transition_filtering_for_a_node(node_indx: int, p_indxs: typing.List, nodes_number: int) \ |
||||
-> np.ndarray: |
||||
"""Builds the necessary structure to filter the desired columns indicated by ``node_indx`` and ``p_indxs`` |
||||
in the dataset. |
||||
This structure will be used in the computation of the state transitions values. |
||||
:param node_indx: the index of the node |
||||
:type node_indx: int |
||||
:param p_indxs: the indexes of the node's parents |
||||
:type p_indxs: List |
||||
:param nodes_number: the total number of nodes in the dataset |
||||
:type nodes_number: int |
||||
:return: The filtering structure for transitions estimation |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
return np.array([node_indx + nodes_number, node_indx, *p_indxs], dtype=np.int) |
||||
|
||||
@staticmethod |
||||
def build_p_comb_structure_for_a_node(parents_values: typing.List) -> np.ndarray: |
||||
""" |
||||
Builds the combinatorial structure that contains the combinations of all the values contained in |
||||
``parents_values``. |
||||
|
||||
:param parents_values: the cardinalities of the nodes |
||||
:type parents_values: List |
||||
:return: A numpy matrix containing a grid of the combinations |
||||
:rtype: numpy.ndArray |
||||
""" |
||||
tmp = [] |
||||
for val in parents_values: |
||||
tmp.append([x for x in range(val)]) |
||||
if len(parents_values) > 0: |
||||
parents_comb = np.array(np.meshgrid(*tmp)).T.reshape(-1, len(parents_values)) |
||||
if len(parents_values) > 1: |
||||
tmp_comb = parents_comb[:, 1].copy() |
||||
parents_comb[:, 1] = parents_comb[:, 0].copy() |
||||
parents_comb[:, 0] = tmp_comb |
||||
else: |
||||
parents_comb = np.array([[]], dtype=np.int) |
||||
return parents_comb |
||||
|
||||
def get_parents_by_id(self, node_id) -> typing.List: |
||||
"""Returns a list of labels of the parents of the node ``node_id`` |
||||
|
||||
:param node_id: the node label |
||||
:type node_id: string |
||||
:return: a List of labels of the parents |
||||
:rtype: List |
||||
""" |
||||
return list(self._graph.predecessors(node_id)) |
||||
|
||||
def get_states_number(self, node_id) -> int: |
||||
return self._graph.nodes[node_id]['val'] |
||||
|
||||
def get_node_indx(self, node_id) -> int: |
||||
return nx.get_node_attributes(self._graph, 'indx')[node_id] |
||||
|
||||
def get_positional_node_indx(self, node_id) -> int: |
||||
return self._graph.nodes[node_id]['pos_indx'] |
||||
|
||||
@property |
||||
def nodes(self) -> typing.List: |
||||
return self._graph_struct.nodes_labels |
||||
|
||||
@property |
||||
def edges(self) -> typing.List: |
||||
return list(self._graph.edges) |
||||
|
||||
@property |
||||
def nodes_indexes(self) -> np.ndarray: |
||||
return self._graph_struct.nodes_indexes |
||||
|
||||
@property |
||||
def nodes_values(self) -> np.ndarray: |
||||
return self._graph_struct.nodes_values |
||||
|
||||
@property |
||||
def time_scalar_indexing_strucure(self) -> np.ndarray: |
||||
return self._time_scalar_indexing_structure |
||||
|
||||
@property |
||||
def time_filtering(self) -> np.ndarray: |
||||
return self._time_filtering |
||||
|
||||
@property |
||||
def transition_scalar_indexing_structure(self) -> np.ndarray: |
||||
return self._transition_scalar_indexing_structure |
||||
|
||||
@property |
||||
def transition_filtering(self) -> np.ndarray: |
||||
return self._transition_filtering |
||||
|
||||
@property |
||||
def p_combs(self) -> np.ndarray: |
||||
return self._p_combs_structure |
||||
@ -1,95 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
|
||||
import numpy as np |
||||
import pandas as pd |
||||
|
||||
import .abstract_sample_path as asam |
||||
import ..utility.json_importer as imp |
||||
from .structure import Structure |
||||
from .trajectory import Trajectory |
||||
import ..utility.abstract_importer as ai |
||||
|
||||
|
||||
|
||||
class SamplePath(object): |
||||
"""Aggregates all the informations about the trajectories, the real structure of the sampled net and variables |
||||
cardinalites. Has the task of creating the objects ``Trajectory`` and ``Structure`` that will |
||||
contain the mentioned data. |
||||
|
||||
:param importer: the Importer object which contains the imported and processed data |
||||
:type importer: AbstractImporter |
||||
:_trajectories: the ``Trajectory`` object that will contain all the concatenated trajectories |
||||
:_structure: the ``Structure`` Object that will contain all the structural infos about the net |
||||
:_total_variables_count: the number of variables in the net |
||||
""" |
||||
def __init__(self, importer: ai.AbstractImporter): |
||||
"""Constructor Method |
||||
""" |
||||
self._importer = importer |
||||
if self._importer._df_variables is None or self._importer._concatenated_samples is None: |
||||
raise RuntimeError('The importer object has to contain the all processed data!') |
||||
if self._importer._df_variables.empty: |
||||
raise RuntimeError('The importer object has to contain the all processed data!') |
||||
if isinstance(self._importer._concatenated_samples, pd.DataFrame): |
||||
if self._importer._concatenated_samples.empty: |
||||
raise RuntimeError('The importer object has to contain the all processed data!') |
||||
if isinstance(self._importer._concatenated_samples, np.ndarray): |
||||
if self._importer._concatenated_samples.size == 0: |
||||
raise RuntimeError('The importer object has to contain the all processed data!') |
||||
self._trajectories = None |
||||
self._structure = None |
||||
self._total_variables_count = None |
||||
|
||||
def build_trajectories(self) -> None: |
||||
"""Builds the Trajectory object that will contain all the trajectories. |
||||
Clears all the unused dataframes in ``_importer`` Object |
||||
""" |
||||
self._trajectories = \ |
||||
Trajectory(self._importer.build_list_of_samples_array(self._importer.concatenated_samples), |
||||
len(self._importer.sorter) + 1) |
||||
self._importer.clear_concatenated_frame() |
||||
|
||||
def build_structure(self) -> None: |
||||
""" |
||||
Builds the ``Structure`` object that aggregates all the infos about the net. |
||||
""" |
||||
if self._importer.sorter != self._importer.variables.iloc[:, 0].to_list(): |
||||
raise RuntimeError("The Dataset columns order have to match the order of labels in the variables Frame!") |
||||
|
||||
self._total_variables_count = len(self._importer.sorter) |
||||
labels = self._importer.variables.iloc[:, 0].to_list() |
||||
indxs = self._importer.variables.index.to_numpy() |
||||
vals = self._importer.variables.iloc[:, 1].to_numpy() |
||||
if self._importer.structure is None or self._importer.structure.empty: |
||||
edges = [] |
||||
else: |
||||
edges = list(self._importer.structure.to_records(index=False)) |
||||
self._structure = Structure(labels, indxs, vals, edges, |
||||
self._total_variables_count) |
||||
|
||||
def clear_memory(self): |
||||
self._importer._raw_data = [] |
||||
|
||||
@property |
||||
def trajectories(self) -> Trajectory: |
||||
return self._trajectories |
||||
|
||||
@property |
||||
def structure(self) -> Structure: |
||||
return self._structure |
||||
|
||||
@property |
||||
def total_variables_count(self) -> int: |
||||
return self._total_variables_count |
||||
|
||||
@property |
||||
def has_prior_net_structure(self) -> bool: |
||||
return bool(self._structure.edges) |
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
@ -1,98 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
import typing |
||||
|
||||
import numpy as np |
||||
|
||||
import structure_graph.conditional_intensity_matrix as cim |
||||
|
||||
|
||||
class SetOfCims(object): |
||||
"""Aggregates all the CIMS of the node identified by the label _node_id. |
||||
|
||||
:param node_id: the node label |
||||
:type node_ind: string |
||||
:param parents_states_number: the cardinalities of the parents |
||||
:type parents_states_number: List |
||||
:param node_states_number: the caridinality of the node |
||||
:type node_states_number: int |
||||
:param p_combs: the p_comb structure bound to this node |
||||
:type p_combs: numpy.ndArray |
||||
:_state_residence_time: matrix containing all the state residence time vectors for the node |
||||
:_transition_matrices: matrix containing all the transition matrices for the node |
||||
:_actual_cims: the cims of the node |
||||
""" |
||||
|
||||
def __init__(self, node_id: str, parents_states_number: typing.List, node_states_number: int, p_combs: np.ndarray): |
||||
"""Constructor Method |
||||
""" |
||||
self._node_id = node_id |
||||
self._parents_states_number = parents_states_number |
||||
self._node_states_number = node_states_number |
||||
self._actual_cims = [] |
||||
self._state_residence_times = None |
||||
self._transition_matrices = None |
||||
self._p_combs = p_combs |
||||
self.build_times_and_transitions_structures() |
||||
|
||||
def build_times_and_transitions_structures(self) -> None: |
||||
"""Initializes at the correct dimensions the state residence times matrix and the state transition matrices. |
||||
""" |
||||
if not self._parents_states_number: |
||||
self._state_residence_times = np.zeros((1, self._node_states_number), dtype=np.float) |
||||
self._transition_matrices = np.zeros((1, self._node_states_number, self._node_states_number), dtype=np.int) |
||||
else: |
||||
self._state_residence_times = \ |
||||
np.zeros((np.prod(self._parents_states_number), self._node_states_number), dtype=np.float) |
||||
self._transition_matrices = np.zeros([np.prod(self._parents_states_number), self._node_states_number, |
||||
self._node_states_number], dtype=np.int) |
||||
|
||||
def build_cims(self, state_res_times: np.ndarray, transition_matrices: np.ndarray) -> None: |
||||
"""Build the ``ConditionalIntensityMatrix`` objects given the state residence times and transitions matrices. |
||||
Compute the cim coefficients.The class member ``_actual_cims`` will contain the computed cims. |
||||
|
||||
:param state_res_times: the state residence times matrix |
||||
:type state_res_times: numpy.ndArray |
||||
:param transition_matrices: the transition matrices |
||||
:type transition_matrices: numpy.ndArray |
||||
""" |
||||
for state_res_time_vector, transition_matrix in zip(state_res_times, transition_matrices): |
||||
cim_to_add = cim.ConditionalIntensityMatrix(state_res_time_vector, transition_matrix) |
||||
cim_to_add.compute_cim_coefficients() |
||||
self._actual_cims.append(cim_to_add) |
||||
self._actual_cims = np.array(self._actual_cims) |
||||
self._transition_matrices = None |
||||
self._state_residence_times = None |
||||
|
||||
def filter_cims_with_mask(self, mask_arr: np.ndarray, comb: typing.List) -> np.ndarray: |
||||
"""Filter the cims contained in the array ``_actual_cims`` given the boolean mask ``mask_arr`` and the index |
||||
``comb``. |
||||
|
||||
:param mask_arr: the boolean mask that indicates which parent to consider |
||||
:type mask_arr: numpy.array |
||||
:param comb: the state/s of the filtered parents |
||||
:type comb: numpy.array |
||||
:return: Array of ``ConditionalIntensityMatrix`` objects |
||||
:rtype: numpy.array |
||||
""" |
||||
if mask_arr.size <= 1: |
||||
return self._actual_cims |
||||
else: |
||||
flat_indxs = np.argwhere(np.all(self._p_combs[:, mask_arr] == comb, axis=1)).ravel() |
||||
return self._actual_cims[flat_indxs] |
||||
|
||||
@property |
||||
def actual_cims(self) -> np.ndarray: |
||||
return self._actual_cims |
||||
|
||||
@property |
||||
def p_combs(self) -> np.ndarray: |
||||
return self._p_combs |
||||
|
||||
def get_cims_number(self): |
||||
return len(self._actual_cims) |
||||
|
||||
|
||||
|
||||
|
||||
@ -1,28 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
import structure_graph.set_of_cims as socim |
||||
|
||||
|
||||
class SetsOfCimsContainer: |
||||
""" |
||||
Aggrega un insieme di oggetti SetOfCims |
||||
""" |
||||
def __init__(self, list_of_keys, states_number_per_node, list_of_parents_states_number, p_combs_list): |
||||
self.sets_of_cims = None |
||||
self.init_cims_structure(list_of_keys, states_number_per_node, list_of_parents_states_number, p_combs_list) |
||||
#self.states_per_variable = states_number |
||||
|
||||
def init_cims_structure(self, keys, states_number_per_node, list_of_parents_states_number, p_combs_list): |
||||
"""for indx, key in enumerate(keys): |
||||
self.sets_of_cims.append( |
||||
socim.SetOfCims(key, list_of_parents_states_number[indx], states_number_per_node[indx]))""" |
||||
self.sets_of_cims = [socim.SetOfCims(pair[1], list_of_parents_states_number[pair[0]], states_number_per_node[pair[0]], p_combs_list[pair[0]]) |
||||
for pair in enumerate(keys)] |
||||
|
||||
def get_set_of_cims(self, node_indx): |
||||
return self.sets_of_cims[node_indx] |
||||
|
||||
def get_cims_of_node(self, node_indx, cim_indx): |
||||
return self.sets_of_cims[node_indx].get_cim(cim_indx) |
||||
|
||||
@ -1,128 +0,0 @@ |
||||
import sys |
||||
sys.path.append('../') |
||||
|
||||
import typing as ty |
||||
|
||||
import numpy as np |
||||
|
||||
|
||||
class Structure(object): |
||||
"""Contains all the infos about the network structure(nodes labels, nodes caridinalites, edges, indexes) |
||||
|
||||
:param nodes_labels_list: the symbolic names of the variables |
||||
:type nodes_labels_list: List |
||||
:param nodes_indexes_arr: the indexes of the nodes |
||||
:type nodes_indexes_arr: numpy.ndArray |
||||
:param nodes_vals_arr: the cardinalites of the nodes |
||||
:type nodes_vals_arr: numpy.ndArray |
||||
:param edges_list: the edges of the network |
||||
:type edges_list: List |
||||
:param total_variables_number: the total number of variables in the dataset |
||||
:type total_variables_number: int |
||||
""" |
||||
|
||||
def __init__(self, nodes_labels_list: ty.List, nodes_indexes_arr: np.ndarray, nodes_vals_arr: np.ndarray, |
||||
edges_list: ty.List, total_variables_number: int): |
||||
"""Constructor Method |
||||
""" |
||||
self._nodes_labels_list = nodes_labels_list |
||||
self._nodes_indexes_arr = nodes_indexes_arr |
||||
self._nodes_vals_arr = nodes_vals_arr |
||||
self._edges_list = edges_list |
||||
self._total_variables_number = total_variables_number |
||||
|
||||
def remove_node(self, node_id: str) -> None: |
||||
"""Remove the node ``node_id`` from all the class members. |
||||
The class member ``_total_variables_number`` since it refers to the total number of variables in the dataset. |
||||
""" |
||||
node_positional_indx = self._nodes_labels_list.index(node_id) |
||||
del self._nodes_labels_list[node_positional_indx] |
||||
self._nodes_indexes_arr = np.delete(self._nodes_indexes_arr, node_positional_indx) |
||||
self._nodes_vals_arr = np.delete(self._nodes_vals_arr, node_positional_indx) |
||||
self._edges_list = [(from_node, to_node) for (from_node, to_node) in self._edges_list if (from_node != node_id |
||||
and to_node != node_id)] |
||||
|
||||
@property |
||||
def edges(self) -> ty.List: |
||||
return self._edges_list |
||||
|
||||
@property |
||||
def nodes_labels(self) -> ty.List: |
||||
return self._nodes_labels_list |
||||
|
||||
@property |
||||
def nodes_indexes(self) -> np.ndarray: |
||||
return self._nodes_indexes_arr |
||||
|
||||
@property |
||||
def nodes_values(self) -> np.ndarray: |
||||
return self._nodes_vals_arr |
||||
|
||||
@property |
||||
def total_variables_number(self) -> int: |
||||
return self._total_variables_number |
||||
|
||||
def get_node_id(self, node_indx: int) -> str: |
||||
""" |
||||
Given the ``node_index`` returns the node label. |
||||
|
||||
:param node_indx: the node index |
||||
:type node_indx: int |
||||
:return: the node label |
||||
:rtype: string |
||||
""" |
||||
return self._nodes_labels_list[node_indx] |
||||
|
||||
def clean_structure_edges(self): |
||||
self._edges_list = list() |
||||
|
||||
def add_edge(self,edge: tuple): |
||||
self._edges_list.append(tuple) |
||||
print(self._edges_list) |
||||
|
||||
def remove_edge(self,edge: tuple): |
||||
self._edges_list.remove(tuple) |
||||
|
||||
def contains_edge(self,edge:tuple) -> bool: |
||||
return edge in self._edges_list |
||||
|
||||
def get_node_indx(self, node_id: str) -> int: |
||||
""" |
||||
Given the ``node_index`` returns the node label. |
||||
|
||||
:param node_id: the node label |
||||
:type node_id: string |
||||
:return: the node index |
||||
:rtype: int |
||||
""" |
||||
pos_indx = self._nodes_labels_list.index(node_id) |
||||
return self._nodes_indexes_arr[pos_indx] |
||||
|
||||
def get_positional_node_indx(self, node_id: str) -> int: |
||||
return self._nodes_labels_list.index(node_id) |
||||
|
||||
def get_states_number(self, node: str) -> int: |
||||
"""Given the node label ``node`` returns the cardinality of the node. |
||||
|
||||
:param node: the node label |
||||
:type node: string |
||||
:return: the node cardinality |
||||
:rtype: int |
||||
""" |
||||
pos_indx = self._nodes_labels_list.index(node) |
||||
return self._nodes_vals_arr[pos_indx] |
||||
|
||||
def __repr__(self): |
||||
return "Variables:\n" + str(self._nodes_labels_list) +"\nValues:\n"+ str(self._nodes_vals_arr) +\ |
||||
"\nEdges: \n" + str(self._edges_list) |
||||
|
||||
def __eq__(self, other): |
||||
"""Overrides the default implementation""" |
||||
if isinstance(other, Structure): |
||||
return set(self._nodes_labels_list) == set(other._nodes_labels_list) and \ |
||||
np.array_equal(self._nodes_vals_arr, other._nodes_vals_arr) and \ |
||||
np.array_equal(self._nodes_indexes_arr, other._nodes_indexes_arr) and \ |
||||
self._edges_list == other._edges_list |
||||
|
||||
return False |
||||
|
||||
@ -1,28 +0,0 @@ |
||||
from functools import wraps |
||||
from time import time |
||||
|
||||
def timing(f): |
||||
@wraps(f) |
||||
def wrap(*args, **kw): |
||||
ts = time() |
||||
result = f(*args, **kw) |
||||
te = time() |
||||
print (f"{f.__name__} args:[{args},{kw}] took: {te-ts} sec") |
||||
|
||||
return result |
||||
return wrap |
||||
|
||||
|
||||
def timing_write(f): |
||||
@wraps(f) |
||||
def wrap(*args, **kw): |
||||
ts = time() |
||||
result = f(*args, **kw) |
||||
te = time() |
||||
print (f"{f.__name__} args:[{args},{kw}] took: {te-ts} sec") |
||||
|
||||
with open("../results/results.csv", 'a+') as fi: |
||||
fi.write(f"\n{round(te-ts,3)},") |
||||
|
||||
return result |
||||
return wrap |
||||
File diff suppressed because it is too large
Load Diff
|
|
File diff suppressed because it is too large
Load Diff
@ -1,147 +0,0 @@ |
||||
import sys |
||||
sys.path.append("../../classes/") |
||||
import glob |
||||
import math |
||||
import os |
||||
import unittest |
||||
|
||||
import networkx as nx |
||||
import numpy as np |
||||
import pandas as pd |
||||
import psutil |
||||
from line_profiler import LineProfiler |
||||
import copy |
||||
import json |
||||
|
||||
import utility.cache as ch |
||||
import structure_graph.sample_path as sp |
||||
import estimators.structure_score_based_estimator as se_score |
||||
import estimators.structure_constraint_based_estimator as se_constr |
||||
import utility.sample_importer as si |
||||
|
||||
|
||||
|
||||
class TestTabuSearch(unittest.TestCase): |
||||
|
||||
@classmethod |
||||
def setUpClass(cls): |
||||
pass |
||||
|
||||
|
||||
|
||||
def test_constr(self): |
||||
|
||||
list_constraint= [0,1] |
||||
|
||||
list_cardinality= [[2,"binary"],[3,"ternary"], [4,"quaternary"]] |
||||
|
||||
|
||||
|
||||
list_dens = [["0.1","_01"],["0.2","_02"], ["0.3",""], ["0.4","_04"] ] |
||||
|
||||
for constr in list_constraint: |
||||
for card in list_cardinality: |
||||
for dens in list_dens: |
||||
|
||||
if card[0] == 4: |
||||
list_vals= [3,4,5,6,10,15] |
||||
else: |
||||
list_vals= [3,4,5,6,10,15,20] |
||||
|
||||
for var_n in list_vals: |
||||
|
||||
patience = 25 |
||||
|
||||
var_number= var_n |
||||
|
||||
if var_number > 11: |
||||
patience = 30 |
||||
|
||||
if var_number > 16: |
||||
patience = 35 |
||||
|
||||
|
||||
cardinality = card[0] |
||||
cardinality_string = card[1] |
||||
|
||||
density= dens[0] |
||||
density_string = dens[1] |
||||
|
||||
constraint = constr |
||||
|
||||
index = 1 |
||||
num_networks=10 |
||||
|
||||
|
||||
while index <= num_networks: |
||||
|
||||
with open(f"/home/alessandro/Documents/ctbn_cba/data/networks_and_trajectories_{cardinality_string}_data{density_string}_{var_number}/{index}.json") as f: |
||||
raw_data = json.load(f) |
||||
|
||||
trajectory_list_raw= raw_data["samples"] |
||||
|
||||
trajectory_list = [pd.DataFrame(sample) for sample in trajectory_list_raw] |
||||
|
||||
variables= pd.DataFrame(raw_data["variables"]) |
||||
prior_net_structure = pd.DataFrame(raw_data["dyn.str"]) |
||||
|
||||
|
||||
self.importer = si.SampleImporter( |
||||
trajectory_list=trajectory_list, |
||||
variables=variables, |
||||
prior_net_structure=prior_net_structure |
||||
) |
||||
|
||||
self.importer.import_data() |
||||
self.s1 = sp.SamplePath(self.importer) |
||||
self.s1.build_trajectories() |
||||
self.s1.build_structure() |
||||
|
||||
|
||||
true_edges = copy.deepcopy(self.s1.structure.edges) |
||||
true_edges = set(map(tuple, true_edges)) |
||||
|
||||
if constr == 0: |
||||
se1 = se_score.StructureScoreBasedEstimator(self.s1) |
||||
set_list_edges = se1.estimate_structure( |
||||
max_parents = None, |
||||
iterations_number = 100, |
||||
patience = patience, |
||||
tabu_length = var_number, |
||||
tabu_rules_duration = var_number, |
||||
optimizer = 'tabu' |
||||
) |
||||
else: |
||||
se1 = se_constr.StructureConstraintBasedEstimator(self.s1,0.1,0.1) |
||||
set_list_edges = se1.estimate_structure(disable_multiprocessing=False) |
||||
|
||||
n_added_fake_edges = len(set_list_edges.difference(true_edges)) |
||||
|
||||
n_missing_edges = len(true_edges.difference(set_list_edges)) |
||||
|
||||
n_true_positive = len(true_edges) - n_missing_edges |
||||
|
||||
precision = n_true_positive / (n_true_positive + n_added_fake_edges) |
||||
|
||||
recall = n_true_positive / (n_true_positive + n_missing_edges) |
||||
|
||||
f1_measure = round(2* (precision*recall) / (precision+recall),3) |
||||
|
||||
print(true_edges) |
||||
print(set_list_edges) |
||||
print(f"precision: {precision} ") |
||||
print(f"recall: {recall} ") |
||||
|
||||
with open("../results/results.csv", 'a+') as fi: |
||||
fi.write(f"{constraint},{var_number},{density},{cardinality},{index},{f1_measure},{round(precision,3)},{round(recall,3)}") |
||||
|
||||
index += 1 |
||||
|
||||
self.assertEqual(set_list_edges, true_edges) |
||||
|
||||
|
||||
|
||||
|
||||
if __name__ == '__main__': |
||||
unittest.main() |
||||
|
||||
@ -1,108 +0,0 @@ |
||||
import sys |
||||
sys.path.append("../../classes/") |
||||
import glob |
||||
import math |
||||
import os |
||||
import unittest |
||||
|
||||
import networkx as nx |
||||
import numpy as np |
||||
import psutil |
||||
from line_profiler import LineProfiler |
||||
import copy |
||||
|
||||
import utility.cache as ch |
||||
import structure_graph.sample_path as sp |
||||
import estimators.structure_constraint_based_estimator as se |
||||
import utility.json_importer as ji |
||||
|
||||
|
||||
|
||||
class TestTabuSearch(unittest.TestCase): |
||||
|
||||
@classmethod |
||||
def setUpClass(cls): |
||||
pass |
||||
|
||||
|
||||
|
||||
def test_constr(self): |
||||
|
||||
list_vals= [3,4,5,6,10,15] |
||||
|
||||
list_dens = [["0.1","_01"],["0.2","_02"], ["0.3",""], ["0.4","_04"] ] |
||||
|
||||
for dens in list_dens: |
||||
for var_n in list_vals: |
||||
|
||||
var_number= var_n |
||||
|
||||
cardinality = 4 |
||||
cardinality_string = "quaternary" |
||||
|
||||
density= dens[0] |
||||
density_string = dens[1] |
||||
|
||||
|
||||
|
||||
constraint = 1 |
||||
|
||||
|
||||
|
||||
index = 0 |
||||
num_networks=10 |
||||
|
||||
if var_number > 9: |
||||
num_networks=3 |
||||
|
||||
while index < num_networks: |
||||
#cls.read_files = glob.glob(os.path.join('../../data', "*.json")) |
||||
self.importer = ji.JsonImporter(f"../../data/networks_and_trajectories_{cardinality_string}_data{density_string}_{var_number}.json", |
||||
'samples', 'dyn.str', 'variables', 'Time', 'Name', index ) |
||||
self.s1 = sp.SamplePath(self.importer) |
||||
self.s1.build_trajectories() |
||||
self.s1.build_structure() |
||||
|
||||
|
||||
true_edges = copy.deepcopy(self.s1.structure.edges) |
||||
true_edges = set(map(tuple, true_edges)) |
||||
|
||||
se1 = se.StructureConstraintBasedEstimator(self.s1, 0.1, 0.1) |
||||
se1.ctpc_algorithm() |
||||
|
||||
|
||||
set_list_edges = set(se1.complete_graph.edges) |
||||
|
||||
|
||||
n_added_fake_edges = len(set_list_edges.difference(true_edges)) |
||||
|
||||
n_missing_edges = len(true_edges.difference(set_list_edges)) |
||||
|
||||
n_true_positive = len(true_edges) - n_missing_edges |
||||
|
||||
precision = n_true_positive / (n_true_positive + n_added_fake_edges) |
||||
|
||||
recall = n_true_positive / (n_true_positive + n_missing_edges) |
||||
|
||||
f1_measure = round(2* (precision*recall) / (precision+recall),3) |
||||
|
||||
# print(f"n archi reali non trovati: {n_missing_edges}") |
||||
# print(f"n archi non reali aggiunti: {n_added_fake_edges}") |
||||
print(true_edges) |
||||
print(set_list_edges) |
||||
print(f"precision: {precision} ") |
||||
print(f"recall: {recall} ") |
||||
|
||||
|
||||
with open("../results/results.csv", 'a+') as fi: |
||||
fi.write(f"{constraint},{var_number},{density},{cardinality},{index},{f1_measure},{round(precision,3)},{round(recall,3)}") |
||||
|
||||
index += 1 |
||||
|
||||
self.assertEqual(set_list_edges, true_edges) |
||||
|
||||
|
||||
|
||||
if __name__ == '__main__': |
||||
unittest.main() |
||||
|
||||
@ -1,122 +0,0 @@ |
||||
import sys |
||||
sys.path.append("../../classes/") |
||||
import glob |
||||
import math |
||||
import os |
||||
import unittest |
||||
|
||||
import networkx as nx |
||||
import numpy as np |
||||
import psutil |
||||
from line_profiler import LineProfiler |
||||
import copy |
||||
|
||||
import utility.cache as ch |
||||
import structure_graph.sample_path as sp |
||||
import estimators.structure_score_based_estimator as se |
||||
import utility.json_importer as ji |
||||
|
||||
|
||||
|
||||
class TestTabuSearch(unittest.TestCase): |
||||
|
||||
@classmethod |
||||
def setUpClass(cls): |
||||
pass |
||||
|
||||
|
||||
|
||||
def test_constr(self): |
||||
|
||||
|
||||
list_vals= [3,4,5,6,10,15] |
||||
|
||||
list_dens = [["0.1","_01"],["0.2","_02"], ["0.3",""], ["0.4","_04"] ] |
||||
|
||||
for dens in list_dens: |
||||
for var_n in list_vals: |
||||
|
||||
patience = 20 |
||||
|
||||
var_number= var_n |
||||
|
||||
if var_number > 11: |
||||
patience = 25 |
||||
|
||||
if var_number > 16: |
||||
patience = 35 |
||||
|
||||
|
||||
cardinality = 4 |
||||
cardinality_string = "quaternary" |
||||
|
||||
density= dens[0] |
||||
density_string = dens[1] |
||||
|
||||
constraint = 0 |
||||
|
||||
index = 0 |
||||
num_networks=10 |
||||
|
||||
if var_number > 9: |
||||
num_networks=3 |
||||
|
||||
|
||||
while index < num_networks: |
||||
#cls.read_files = glob.glob(os.path.join('../../data', "*.json")) |
||||
self.importer = ji.JsonImporter(f"../../data/networks_and_trajectories_{cardinality_string}_data{density_string}_{var_number}.json", |
||||
'samples', 'dyn.str', 'variables', 'Time', 'Name', index ) |
||||
self.s1 = sp.SamplePath(self.importer) |
||||
self.s1.build_trajectories() |
||||
self.s1.build_structure() |
||||
|
||||
|
||||
true_edges = copy.deepcopy(self.s1.structure.edges) |
||||
true_edges = set(map(tuple, true_edges)) |
||||
|
||||
se1 = se.StructureScoreBasedEstimator(self.s1) |
||||
set_list_edges = se1.estimate_structure( |
||||
max_parents = None, |
||||
iterations_number = 100, |
||||
patience = patience, |
||||
tabu_length = var_number, |
||||
tabu_rules_duration = var_number, |
||||
optimizer = 'hill' |
||||
) |
||||
|
||||
|
||||
|
||||
|
||||
|
||||
n_added_fake_edges = len(set_list_edges.difference(true_edges)) |
||||
|
||||
n_missing_edges = len(true_edges.difference(set_list_edges)) |
||||
|
||||
n_true_positive = len(true_edges) - n_missing_edges |
||||
|
||||
precision = n_true_positive / (n_true_positive + n_added_fake_edges) |
||||
|
||||
recall = n_true_positive / (n_true_positive + n_missing_edges) |
||||
|
||||
f1_measure = round(2* (precision*recall) / (precision+recall),3) |
||||
|
||||
# print(f"n archi reali non trovati: {n_missing_edges}") |
||||
# print(f"n archi non reali aggiunti: {n_added_fake_edges}") |
||||
print(true_edges) |
||||
print(set_list_edges) |
||||
print(f"precision: {precision} ") |
||||
print(f"recall: {recall} ") |
||||
|
||||
with open("../results/results.csv", 'a+') as fi: |
||||
fi.write(f"{constraint},{var_number},{density},{cardinality},{index},{f1_measure},{round(precision,3)},{round(recall,3)}") |
||||
|
||||
index += 1 |
||||
|
||||
self.assertEqual(set_list_edges, true_edges) |
||||
|
||||
|
||||
|
||||
|
||||
if __name__ == '__main__': |
||||
unittest.main() |
||||
|
||||
Reference in new issue