def __init__(
self,
n_clusters=8,
init='random',
n_init=1,
max_iter=300,
tol=1e-4,
precompute_distances='auto',
verbose=0,
random_state=None,
copy_x=True,
n_jobs=1,
algorithm='auto',
# Beyond sklearn (with optimal defaults)
gpu_id=0,
n_gpus=-1,
init_data="randomselect",
do_checks=1,
backend='auto'):
import os
_backend = os.environ.get('H2O4GPU_BACKEND', None)
if _backend is not None:
backend = _backend
assert_is_type(backend, str)
# FIXME: Add init as array and kmeans++ to h2o4gpu
# setup backup to sklearn class
# (can remove if fully implement sklearn functionality)
self.do_sklearn = False
if backend == 'auto':
example = np.array([1, 2, 3])
# pylint: disable=unidiomatic-typecheck
if type(init) == type(example):
print("WARNING: init as ndarray of centers not yet supported."
" Using sklearn.")
self.do_sklearn = True
else:
if init == "k-means++":
print("WARNING: init as k-means++ not yet supported."
" Using sklearn.")
self.do_sklearn = True
# FIXME: Add n_init to h2o4gpu
if n_init != 1:
print("WARNING: n_init not supported currently."
" Still using h2o4gpu.")
if precompute_distances != "auto":
print("WARNING: precompute_distances not used."
" Still using h2o4gpu.")
elif backend == 'sklearn':
self.do_sklearn = True
elif backend == 'h2o4gpu':
self.do_sklearn = False
if self.do_sklearn:
self.backend = 'sklearn'
else:
self.backend = 'h2o4gpu'
from h2o4gpu.cluster import k_means_
self.model_sklearn = k_means_.KMeansSklearn(
n_clusters=n_clusters,
init=init,
n_init=n_init,
max_iter=max_iter,
tol=tol,
precompute_distances=precompute_distances,
verbose=verbose,
random_state=random_state,
copy_x=copy_x,
n_jobs=n_jobs,
algorithm=algorithm)
self.model_h2o4gpu = KMeansH2O(
n_clusters=n_clusters,
init=init,
n_init=n_init,
max_iter=max_iter,
tol=tol,
precompute_distances=precompute_distances,
verbose=verbose,
random_state=random_state,
copy_x=copy_x,
n_jobs=n_jobs,
algorithm=algorithm,
# H2O4GPU
gpu_id=gpu_id,
n_gpus=n_gpus,
init_data=init_data,
do_checks=do_checks)
if self.do_sklearn:
self.model = self.model_sklearn
else:
self.model = self.model_h2o4gpu
def __init__(
self,
n_clusters=8,
init='k-means++',
n_init=1,
max_iter=300,
tol=1e-4,
precompute_distances='auto',
verbose=0,
random_state=None,
copy_x=True,
n_jobs=1,
algorithm='auto',
# Beyond sklearn (with optimal defaults)
gpu_id=0,
n_gpus=-1,
do_checks=1,
backend='auto'):
import os
_backend = os.environ.get('H2O4GPU_BACKEND', None)
if _backend is not None:
backend = _backend
# FIXME: Add init as array and kmeans++ to h2o4gpu
# setup backup to sklearn class
# (can remove if fully implement sklearn functionality)
self.do_sklearn = False
if backend == 'auto':
example = np.array([1, 2, 3])
# pylint: disable=unidiomatic-typecheck
if type(init) == type(example):
KMeans._print_verbose(
verbose, 0,
"'init' as ndarray of centers not yet supported."
"Running ScikitLearn CPU version.")
self.do_sklearn = True
# FIXME: Add n_init to h2o4gpu
if n_init != 1:
KMeans._print_verbose(
verbose, 0, "'n_init' not supported. "
"Running h2o4gpu with n_init = 1.")
if precompute_distances != "auto":
KMeans._print_verbose(verbose, 0,
"'precompute_distances' not used.")
elif backend == 'sklearn':
self.do_sklearn = True
elif backend == 'h2o4gpu':
self.do_sklearn = False
if self.do_sklearn:
self.backend = 'sklearn'
else:
self.backend = 'h2o4gpu'
from h2o4gpu.cluster import k_means_
self.model_sklearn = k_means_.KMeansSklearn(
n_clusters=n_clusters,
init=init,
n_init=n_init,
max_iter=max_iter,
tol=tol,
precompute_distances=precompute_distances,
verbose=verbose,
random_state=random_state,
copy_x=copy_x,
n_jobs=n_jobs,
algorithm=algorithm)
self.model_h2o4gpu = KMeansH2O(
n_clusters=n_clusters,
init=init,
n_init=n_init,
max_iter=max_iter,
tol=tol,
precompute_distances=precompute_distances,
verbose=verbose,
random_state=random_state,
copy_x=copy_x,
n_jobs=n_jobs,
algorithm=algorithm,
# H2O4GPU
gpu_id=gpu_id,
n_gpus=n_gpus,
do_checks=do_checks)
# pylint: disable=protected-access
if self.do_sklearn or self.model_h2o4gpu._load_lib() is None:
self.model = self.model_sklearn
KMeans._print_verbose(verbose, 0, "Using ScikitLearn backend.")
else:
self.model = self.model_h2o4gpu
KMeans._print_verbose(verbose, 0, "Using h2o4gpu backend.")