import numpy as np
import tables
import numba as nb
import MyML.utils.profiling as myProf
import MyML.helper.partition as part
import MyML.EAC.sparse as eacSp
import MyML.EAC.eac_new as myEAC
import shutil
t = myProf.Timer()
# ensemble_file = '/home/diogoaos/QCThesis/datasets/gauss10e6_overlap/ensemble_500k_test2.h5'
ensemble_file = '/media/Data/diogoaos_tmp/gaussseparated_ensembles/ensemble_500000_2sqrt.hdf'
coassc_path_ssd = '/home/diogoaos/QCThesis/coassoc.h5'
index_path_ssd = '/home/diogoaos/QCThesis/'
coassc_path_spin = '/media/Data/diogoaos_tmp/coassoc.h5'
index_path_spin = '/media/Data/diogoaos_tmp/'
print "loading ensemble"
t.reset()
t.tic()
ensemble = part.loadEnsembleFromFileHDF(ensemble_file)
print 'load ensemble time: {}'.format(t.tac())
n_samples = part.n_samples_from_partition(ensemble[0])
n_partitions = len(ensemble)
print "number of samples: {}".format(n_samples)
print "number of partitions: {}".format(n_partitions)
import MyML.utils.profiling as myProf
tImport = myProf.Timer()
tImport.tic()
import numpy as np
import MyML.helper.partition as part
import MyML.cluster.K_Means3 as myKM
import MyML.metrics.accuracy as myAcc
import MyML.EAC.eac_new as myEAC
import MyML.EAC.sparse as mySpEAC
import gc
import argparse
import os.path
def correspond(l0, l1):
"""gets two labels arrays and, if they have the
same number of clusters, tries to equal the label assignment
"""
l0_unique = np.unique(l0)
if l0_unique.size != np.unique(l1).size:
return -1
inc = l0_unique.max() + 100
# increment all labels
for l in l0_unique:
l0[l0 == l] = l + inc
print "Path is not directory: ", folder
sys.exit(1)
# await confirmation
if not args.yes:
raw_input("Folder: {}\nIs this correct?".format(folder))
else:
print "Folder being used is: {}".format(folder)
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
# IMPORTS
# # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #
import MyML.utils.profiling as myProf
t = myProf.Timer()
t.tic()
import numpy as np
from numba import cuda
import pandas as pd # for storing results
from sklearn import datasets # generate gaussian mixture
import MyML.cluster.K_Means3 as myKM
myKM.CUDA_PPT = PPT
# Setup logging
import logging
'min_degree', 'max_degree', 'mean_degree', 'std_degree',
'accuracy', 'sl_clusts',
'round','disk']
type_mats = ["full",
"full condensed",
"sparse complete",
"sparse condensed const",
"sparse condensed linear"]
rounds = 1
res_lines = rounds * len(cardinality) * len(rules) * len(type_mats)
results = pd.DataFrame(index=range(res_lines), columns=res_cols)
t = myProf.Timer() # timer
# ensemble properties
n_partitions = 100
n_iters = 3
# EAC properties
sparse_max_assocs_factor = 3
# ## run
logger.info("Starting experiment...")
# In[198]:
res_idx = 0