Skip to content

Commit

Permalink
fix and clean up functions in nodal_roles.py
Browse files Browse the repository at this point in the history
  • Loading branch information
@klbegany
klbegany committed Nov 20, 2015
1 parent 73607dd commit dc8e306
Showing 1 changed file with 65 additions and 59 deletions.
124 changes: 65 additions & 59 deletions brainx/nodal_roles.py
View file
Expand Up @@ -43,6 +43,7 @@ def within_community_strength(weighted_partition, calc_type, edgeless = np.nan,
'''
wc_dict = {}
for c, community in enumerate(weighted_partition.communities):
nodes = weighted_partition.graph.nodes()
community_strengths = []
community_wc_dict = {}
for node in community: #get average within-community strength (i.e. weighted degree)
Expand All @@ -53,28 +54,25 @@ def within_community_strength(weighted_partition, calc_type, edgeless = np.nan,
node_strength = weighted_partition.node_negative_strength(node)
community_strengths.append(weighted_partition.node_negative_strength_by_community(node)[c])
elif calc_type == 'smp':
pos_wc_dict = within_community_strength(weighted_partition, calc_type='pos')
neg_wc_dict = within_community_strength(weighted_partition, calc_type='neg')
keys = pos_wc_dict.keys()
vals = pos_wc_dict.values() - neg_wc_dict.values()
return dict(zip(keys,vals))
pos_wc = np.array(within_community_strength(weighted_partition, calc_type='pos').values())
neg_wc = np.array(within_community_strength(weighted_partition, calc_type='neg').values())
wc = pos_wc - neg_wc
return dict(zip(nodes, wc))
elif calc_type == 'sta':
pos_wc_dict = within_community_strength(weighted_partition, calc_type='pos')
neg_wc_dict = within_community_strength(weighted_partition, calc_type='neg')
keys = pos_wc_dict.keys()
pos_wc = np.array(within_community_strength(weighted_partition, calc_type='pos').values())
neg_wc = np.array(within_community_strength(weighted_partition, calc_type='neg').values())
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
vals = pos_wc_dict.values - (neg_wc_dict.values * (neg_m2 / pos_m2 + neg_m2))
return dict(zip(keys,vals))
wc = pos_wc - neg_wc * (neg_m2 / (pos_m2 + neg_m2))
return dict(zip(nodes, wc))
elif calc_type == 'gja':
pos_wc_dict = within_community_strength(weighted_partition, calc_type='pos')
neg_wc_dict = within_community_strength(weighted_partition, calc_type='neg')
keys = pos_wc_dict.keys()
pos_wc = np.array(within_community_strength(weighted_partition, calc_type='pos').values())
neg_wc = np.array(within_community_strength(weighted_partition, calc_type='neg').values())
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
vals = (pos_wc_dict.values * (pos_m2 / (pos_m2 + neg_m2))) \
- (neg_wc_dict.values * (neg_m2 / (pos_m2 + neg_m2)))
return dict(zip(keys,vals))
wc = pos_wc * (pos_m2 / (pos_m2 + neg_m2)) \
- neg_wc * (neg_m2 / (pos_m2 + neg_m2))
return dict(zip(nodes, wc))
else:
raise ValueError('%s not supported.' % (calc_type))
if node_strength == 0.0: #catch edgeless nodes, this shouldn't count towards avg wcd
Expand Down Expand Up @@ -128,35 +126,33 @@ def participation_coefficient(weighted_partition, calc_type, edgeless =np.nan, c
Dictionary of the participation coefficient of each node.
'''
pc_dict = {}
for node in weighted_partition.graph:
nodes = weighted_partition.graph.nodes()
for node in nodes:
if calc_type == 'pos':
node_strength = weighted_partition.node_positive_strength(node)
node_community_strengths = weighted_partition.node_positive_strength_by_community(node)
elif calc_type == 'neg':
node_strength = weighted_partition.node_negative_strength(node)
node_community_strengths = weighted_partition.node_negative_strength_by_community(node)
elif calc_type == 'smp':
pos_pc = participation_coefficient(weighted_partition, calc_type='pos')
neg_pc = participation_coefficient(weighted_partition, calc_type='neg')
nodes = pos_pc.keys()
pc = pos_pc.values() - neg_pc.values()
pos_pc = np.array(participation_coefficient(weighted_partition, calc_type='pos').values())
neg_pc = np.array(participation_coefficient(weighted_partition, calc_type='neg').values())
pc = pos_pc - neg_pc
return dict(zip(nodes, pc))
elif calc_type == 'sta':
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
pos_pc = participation_coefficient(weighted_partition, calc_type='pos')
neg_pc = participation_coefficient(weighted_partition, calc_type='neg')
nodes = pos_pc.keys()
pc = pos_pc.values() - ((neg_m2 / (pos_m2 + neg_m2)) * neg_pc.values())
pos_pc = np.array(participation_coefficient(weighted_partition, calc_type='pos').values())
neg_pc = np.array(participation_coefficient(weighted_partition, calc_type='neg').values())
pc = pos_pc - (neg_m2 / (pos_m2 + neg_m2)) * neg_pc
return dict(zip(nodes, pc))
elif calc_type == 'gja':
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
pos_pc = participation_coefficient(weighted_partition, calc_type='pos')
neg_pc = participation_coefficient(weighted_partition, calc_type='neg')
nodes = pos_pc.keys()
pc = ((pos_m2 / (pos_m2 + neg_m2)) * pos_pc.values()) \
- ((neg_m2 / (pos_m2 + neg_m2)) * neg_pc.values())
pos_pc = np.array(participation_coefficient(weighted_partition, calc_type='pos').values())
neg_pc = np.array(participation_coefficient(weighted_partition, calc_type='neg').values())
pc = (pos_m2 / (pos_m2 + neg_m2)) * pos_pc \
- (neg_m2 / (pos_m2 + neg_m2)) * neg_pc
return dict(zip(nodes, pc))
else:
raise ValueError('%s not supported.' % (calc_type))
Expand Down Expand Up @@ -206,32 +202,32 @@ def connection_strength(weighted_partition, calc_type):
if calc_type == 'pos':
strengths = []
for node in nodes:
strengths.append((1/n-1)*node_positive_strength(node))
strengths.append((1. / (n - 1)) * weighted_partition.node_positive_strength(node))
return dict(zip(nodes, strengths))
elif calc_type == 'neg':
strengths = []
for node in nodes:
strengths.append((1/n-1)*node_negative_strength(node))
strengths.append((1. / (n - 1)) * weighted_partition.node_negative_strength(node))
return dict(zip(nodes, strengths))
elif calc_type == 'smp':
pos_strengths = connection_strength(weighted_partition, calc_type='pos')
neg_strengths = connection_strength(weighted_partition, calc_type='neg')
pos_strengths = np.array(connection_strength(weighted_partition, calc_type='pos').values())
neg_strengths = np.array(connection_strength(weighted_partition, calc_type='neg').values())
return dict(zip(nodes, pos_strengths - neg_strengths))
elif calc_type == 'sta':
pos_strengths = connection_strength(weighted_partition, calc_type='pos')
neg_strengths = connection_strength(weighted_partition, calc_type='neg')
pos_strengths = np.array(connection_strength(weighted_partition, calc_type='pos').values())
neg_strengths = np.array(connection_strength(weighted_partition, calc_type='neg').values())
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
strengths = pos_strengths - ((neg_m2 / (pos_m2 + neg_m2)) * neg_strenghts)
return dict(zip(nodes, strenghts))
strengths = pos_strengths - (neg_m2 / (pos_m2 + neg_m2)) * neg_strengths
return dict(zip(nodes, strengths))
elif calc_type == 'gja':
pos_strengths = connection_strength(weighted_partition, calc_type='pos')
neg_strengths = connection_strength(weighted_partition, calc_type='neg')
pos_strengths = np.array(connection_strength(weighted_partition, calc_type='pos').values())
neg_strengths = np.array(connection_strength(weighted_partition, calc_type='neg').values())
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
strengths = ((pos_m2 / (pos_m2 + neg_m2)) * pos_strengths) \
- ((neg_m2 / (pos_m2 + neg_m2)) * neg_strenghts)
return dict(zip(nodes, strenghts))
strengths = (pos_m2 / (pos_m2 + neg_m2)) * pos_strengths \
- (neg_m2 / (pos_m2 + neg_m2)) * neg_strengths
return dict(zip(nodes, strengths))
else:
raise ValueError('%s not supported.' % (calc_type))

Expand Down Expand Up @@ -266,39 +262,49 @@ def connection_diversity(weighted_partition, calc_type):
diversity = []
for node in nodes:
node_strength = weighted_partition.node_positive_strength(node)
node_community_strength = weighted_partition.node_positive_strength_by_community(node)
node_diversity = -(1 / np.log(m)) * \
sum(node_community_strength * (np.log(node_community_strength)))
node_community_strength = np.array(weighted_partition.node_positive_strength_by_community(node))
mask = node_community_strength == 0.
node_community_strength = node_community_strength[~mask]# remove zero values
p_i = node_community_strength / node_strength
try:
node_diversity = -(1 / np.log(m)) * sum(p_i * (np.log(p_i)))
except:
node_diversity = 0.
diversity.append(node_diversity)
return dict(zip(nodes, diversity))
elif calc_type == 'neg':
diversity = []
for node in nodes:
node_strength = weighted_partition.node_negative_strength(node)
node_community_strength = weighted_partition.node_negative_strength_by_community(node)
node_diversity = -(1 / np.log(m)) * \
sum(node_community_strength * (np.log(node_community_strength)))
node_community_strength = np.array(weighted_partition.node_negative_strength_by_community(node))
mask = node_community_strength == 0.
node_community_strength = node_community_strength[~mask]# remove zero values
p_i = node_community_strength / node_strength
try:
node_diversity = -(1 / np.log(m)) * sum(p_i * (np.log(p_i)))
except:
node_diversity = 0.
diversity.append(node_diversity)
return dict(zip(nodes, diversity))
elif calc_type == 'smp':
pos_diversity = connection_diversity(weighted_partition, 'pos')
neg_diversity = connection_diversity(weighted_partition, 'neg')
diversity = pos_diversity.values() - neg_diversity.values()
pos_diversity = np.array(connection_diversity(weighted_partition, 'pos').values())
neg_diversity = np.array(connection_diversity(weighted_partition, 'neg').values())
diversity = pos_diversity - neg_diversity
return dict(zip(nodes, diversity))
elif calc_type == 'sta':
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
pos_diversity = connection_diversity(weighted_partition, 'pos')
neg_diversity = connection_diversity(weighted_partition, 'neg')
diversity = pos_diversity.values() - ((neg_m2 / pos_m2 + neg_m2) * neg_diversity.values())
pos_diversity = np.array(connection_diversity(weighted_partition, 'pos').values())
neg_diversity = np.array(connection_diversity(weighted_partition, 'neg').values())
diversity = pos_diversity - (neg_m2 / (pos_m2 + neg_m2)) * neg_diversity
return dict(zip(nodes, diversity))
elif calc_type == 'gja':
pos_m2 = weighted_partition.total_positive_strength
neg_m2 = weighted_partition.total_negative_strength
pos_diversity = connection_diversity(weighted_partition, 'pos')
neg_diversity = connection_diversity(weighted_partition, 'neg')
diversity = ((pos_m2 / pos_m2 + neg_m2) * pos_diversity.values() \
- ((neg_m2 / pos_m2 + neg_m2) * neg_diversity.values())
pos_diversity = np.array(connection_diversity(weighted_partition, 'pos').values())
neg_diversity = np.array(connection_diversity(weighted_partition, 'neg').values())
diversity = (pos_m2 / (pos_m2 + neg_m2)) * pos_diversity \
- (neg_m2 / (pos_m2 + neg_m2)) * neg_diversity
return dict(zip(nodes, diversity))
else:
raise ValueError('%s not supported.' % (calc_type))
Expand Down

0 comments on commit dc8e306

Please sign in to comment.