def strip_small_categories(glyphs):
gc = knn.glyphs_by_category(glyphs)
new_glyphs = []
for v in gc.itervalues():
if len(v) > 3:
new_glyphs.extend(v)
return new_glyphs
def strip_small_categories(glyphs):
gc = knn.glyphs_by_category(glyphs)
new_glyphs = []
for v in gc.values():
if len(v) > 3:
new_glyphs.extend(v)
return new_glyphs
def get_graph_stats(glyphs, k):
s = { }
gc = knn.glyphs_by_category(glyphs)
for x in gc.itervalues():
if len(x) == 1:
s[x[0].get_main_id()] = 1.0
continue
graph = cluster.make_spanning_tree(x, k)
edges = []
for edge in graph.get_edges():
edges.append(edge.cost)
id = x[0].get_main_id()
s[id] = stats.mean(edges)
#cluster.make_subtrees_stddev(graph, 1, 2, 0)
#print graph.nedges, len(x)
return s
def get_graph_stats(glyphs, k):
s = {}
gc = knn.glyphs_by_category(glyphs)
for x in gc.values():
if len(x) == 1:
s[x[0].get_main_id()] = 1.0
continue
graph = cluster.make_spanning_tree(x, k)
edges = []
for edge in graph.get_edges():
edges.append(edge.cost)
id = x[0].get_main_id()
s[id] = stats.mean(edges)
#cluster.make_subtrees_stddev(graph, 1, 2, 0)
#print graph.nedges, len(x)
return s
def cluster2(glyphs):
ko = knn.kNNInterative()
gc = knn.glyphs_by_category(cluster(glyphs, k=ko))
small = []
large = []
for x in gc.itervalues():
if len(x) < 10:
small.extend(x)
else:
large.append(x)
print len(small)
output = cluster(small, 1, 1, label="cluster_small.", k=ko)
cur_label = 0
for x in large:
l = x[0].get_main_id() + ".cluster_large" + str(cur_label) + "."
cur_label += 1
c = cluster(x, .6, 4, label=l, k=ko)
output.extend(c)
return output
