def generate_candidates(k_freq_itemsets, constraint, items):
"""
Returns set of candidates with k + 1 nodes using set of candidates with k nodes and also returns support data which contains the parent child relationship.
:param k_freq_itemsets: List of frequent itemsets of size k
:param constraint: function which returns boolean value
:param items: not used by definition of antimonotone but used by loosely antimonotone definition
:return: List of candidates of size k+1, dictionary of support data with child -> parent mapping
"""
candidates = set()
support_data = {}
for a in k_freq_itemsets:
for b in k_freq_itemsets:
k = len(a)
union = tuple(sorted(set(a).union(set(b))))
if len(union) == k + 1 and constraint(list(union)):
if union not in candidates:
parents = LevelwiseApriori.find_subsets(list(union), k)
if isSubSet(parents, k_freq_itemsets):
candidates.add(union)
if union not in support_data:
support_data[tuple(union)] = parents
# returning set of candidates with k + 1 nodes
return list(candidates), support_data
def generate_candidates(k_freq_itemsets, constraint, items):
"""
Returns set of candidates with k + 1 nodes using set of candidates with k nodes and also returns support data which contains the parent child relationship.
:param k_freq_itemsets: List of frequent itemsets of size k
:param constraint: function which returns boolean value
:param items: not used by definition of antimonotone but used by loosely antimonotone definition
:return: List of candidates of size k+1, dictionary of support data with child -> parent mapping
"""
candidates = set()
support_data = {}
for a in k_freq_itemsets:
for b in k_freq_itemsets:
k = len(a)
union = tuple(sorted(set(a).union(set(b))))
if len(union) == k + 1 and constraint(list(union)):
if union not in candidates:
parents = LevelwiseApriori.find_subsets(list(union), k)
if isSubSet(parents, k_freq_itemsets):
candidates.add(union)
if union not in support_data:
support_data[tuple(union)] = parents
# returning set of candidates with k + 1 nodes
return list(candidates), support_data
def maximal_lam_sigma_ssd_ucs(self, sigma):
return LevelwiseApriori.maximal_freq_itemsets(lambda x: self.compute_scaled_subgraph_divergence(x) <= sigma,
self.nodes(), LooselyAntiMonotone.generate_candidates)
def phi_sd_ucs(self, phi):
return LevelwiseApriori.freq_itemsets((lambda x: self.compute_subgraph_divergence(x) <= phi), self.nodes(),
AntiMonotone.generate_candidates)
