def dtree(dm, method='a'):
"""
dm: DistMat(N), method: s ('a') -> Tree(N)
calls PyCluster.treecluster(method=method, distancematrix=dm), except that
dm is copied, to protect from PyClusters INSANE behaviour of modifying dm
in place.
"""
#warning: pycluster writes the dm by side effect!!!!!
dm = dm.copy()
return treecluster(method=str(method), distancematrix=dm)
def _guide_tree(self):
"""
Build a guide tree from the distance matrix
:author: Woon Wai Keen
:author: Vladimir Likic
"""
n = len(self.dist_matrix)
print(f" -> Clustering {n * (n - 1):d} pairwise alignments.", end='')
self.tree = treecluster(data=None, distancematrix=self.dist_matrix, method='a')
print("Done")
def vtree(dat, dist='e', method='a'):
"""
dat: N,M-#, dist: s ('e'), method: s ('a') -> Tree(N)
calls PyCluster.treecluster(dat, dist=dist, method=method),
This require that dat contain data smaples in rows, and that the distance
function is appropriate to use on the samples (e.g. the default dist='e' is
the euclidean distance, and all PyCluster dist options are vector space
distances).
"""
#implicit transpose=0, aka samples are rows, and there is no mask or weight
return treecluster(dat, dist=dist, method=str(method))
def _guide_tree(self, dist_matrix):
"""
@summary: Build a guide tree from the distance matrix
@param dist_matrix: The distance matrix
@type dist_matrix: numpy.ndarray
@return: Pycluster similarity tree
@rtype: Pycluster.cluster.Tree
@author: Woon Wai Keen
@author: Vladimir Likic
"""
n = len(dist_matrix)
print " -> Clustering %d pairwise alignments." % (n * (n - 1)),
tree = treecluster(distancematrix=dist_matrix, method='a')
print "Done"
return tree
def _guide_tree(self, dist_matrix):
"""
@summary: Build a guide tree from the distance matrix
@param dist_matrix: The distance matrix
@type dist_matrix: numpy.ndarray
@return: Pycluster similarity tree
@rtype: Pycluster.cluster.Tree
@author: Woon Wai Keen
@author: Vladimir Likic
"""
n = len(dist_matrix)
print " -> Clustering %d pairwise alignments." % (n*(n-1)),
tree = treecluster(distancematrix=dist_matrix, method='a')
print "Done"
return tree
print os.getcwd()
im = Image.open(file)
im = im.convert("1")
# Ermittle Bildgroesse:
M = im.size[0]
N = im.size[1]
data = array(im.getdata())
data = data.reshape(im.size)
px1 = vstack([nonzero(data==0)[0],nonzero(data==0)[1]]).T
tree = treecluster(px1)
ho = tree.cut(4)
# Bestimme Zentren:
center = zeros([4,2])
for i in arange(4):
tmp = px1[nonzero(ho==i)[0],:]
center[i,0]=mean(tmp[:,0])
center[i,1]=mean(tmp[:,1])
tmp_ref = tmp-center[i,:]
max_abs = max(sqrt((tmp_ref**2).sum(axis=1))) #bestimme den maximalen Abstand der zum Cluster zugehoerigen Pixel
print max_abs
# Plotte die gefundenen Cluster
def compute_linkage(cls, conden_dist_mat):
lower_dist_mat = cls.convert_condensed_to_lower(conden_dist_mat)
tree = treecluster(distancematrix=lower_dist_mat, method='s')
return cls.convert_tree_to_mat(tree)
#this method generates tree from PyCluster Tree
def generateTree(tree, wordlist):
nodeList = []
for i in range(len(tree)):
parent = Tree()
parent.name = str(-(i+1))
for node in eval(str(tree[i]).split(':')[0]):
if node >= 0:
child = Tree()
child.name = wordlist[node]
parent.add_child(child)
else:
parent.add_child(nodeList[int(-node)-1])
nodeList.append(parent)
print nodeList[-1].get_ascii(show_internal=True)
return nodeList[-1]
if __name__ == "__main__":
db = db_connection['PW_test']
#tree = buildCategoryForest("Travel")
#calDistanceMatrix(wordlist1, tree)
wordlist1 = generateWordList('Travel', 50)
for speech in ['n', 'v']:
d = simToDistMatrix(calSimMatrix2(wordlist1[speech], speech, 'Travel'))
tree = treecluster(method = 'a', dist = 'e', distancematrix = d)
generateTree(tree, wordlist1[speech])
print tree
for i in range(len(wordlist1[speech])):
print i,wordlist1[speech][i]
