def evaluate_sys(run_id, task_type, mth, dataset, ens_method, enable_meta,
eval_type='holdout', time_limit=1200, seed=1, tree_id=0):
_task_type = MULTICLASS_CLS if task_type == 'cls' else REGRESSION
train_data, test_data = load_train_test_data(dataset, task_type=_task_type)
_enable_meta = True if enable_meta == 'true' else False
if task_type == 'cls':
from mindware.estimators import Classifier
estimator = Classifier(time_limit=time_limit,
per_run_time_limit=30,
output_dir=save_folder,
ensemble_method=ens_method,
enable_meta_algorithm_selection=_enable_meta,
evaluation=eval_type,
metric='bal_acc',
include_algorithms=['extra_trees', 'random_forest',
'adaboost', 'gradient_boosting',
'k_nearest_neighbors', 'liblinear_svc',
'libsvm_svc', 'lightgbm',
'logistic_regression', 'random_forest'],
n_jobs=1)
else:
from mindware.estimators import Regressor
estimator = Regressor(time_limit=time_limit,
per_run_time_limit=90,
output_dir=save_folder,
ensemble_method=ens_method,
enable_meta_algorithm_selection=_enable_meta,
evaluation=eval_type,
metric='mse',
# include_preprocessors=['percentile_selector_regression'],
# include_algorithms=['random_forest'],
n_jobs=1)
start_time = time.time()
estimator.fit(train_data, opt_strategy=mth, dataset_id=dataset, tree_id=tree_id)
pred = estimator.predict(test_data)
if task_type == 'cls':
test_score = balanced_accuracy_score(test_data.data[1], pred)
else:
test_score = mean_squared_error(test_data.data[1], pred)
validation_score = estimator._ml_engine.solver.incumbent_perf
# eval_dict = estimator._ml_engine.solver.get_eval_dict()
print('Run ID : %d' % run_id)
print('Dataset : %s' % dataset)
print('Val/Test score : %f - %f' % (validation_score, test_score))
save_path = save_folder + '%s_%s_%s_%s_%d_%d_%d_%d.pkl' % (
task_type, mth, dataset, enable_meta, time_limit, (ens_method is None), tree_id, run_id)
with open(save_path, 'wb') as f:
pickle.dump([dataset, validation_score, test_score, start_time], f)
# Delete output dir
shutil.rmtree(os.path.join(estimator.get_output_dir()))
def evaluate_hmab(algorithms,
run_id,
time_limit=600,
dataset='credit',
eval_type='holdout',
enable_ens=True,
seed=1):
print('%s\nDataset: %s, Run_id: %d, Budget: %d.\n%s' %
('=' * 50, dataset, run_id, time_limit, '=' * 50))
task_id = '[%s][%s-%d-%d]' % (hmab_id, dataset, len(algorithms),
time_limit)
_start_time = time.time()
train_data, test_data = load_train_test_data(dataset,
task_type=MULTICLASS_CLS)
if enable_ens is True:
ensemble_method = 'ensemble_selection'
else:
ensemble_method = None
clf = Classifier(time_limit=time_limit,
per_run_time_limit=per_run_time_limit,
include_algorithms=algorithms,
amount_of_resource=None,
output_dir=save_dir,
ensemble_method=ensemble_method,
evaluation=eval_type,
metric='bal_acc',
n_jobs=1)
# clf.fit(train_data, meta_datasets=holdout_datasets)
# clf.fit(train_data, opt_strategy='combined')
clf.fit(train_data)
clf.refit()
pred = clf.predict(test_data)
test_score = balanced_accuracy_score(test_data.data[1], pred)
timestamps, perfs = clf.get_val_stats()
validation_score = np.max(perfs)
print('Dataset : %s' % dataset)
print('Validation/Test score : %f - %f' % (validation_score, test_score))
save_path = save_dir + '%s-%d.pkl' % (task_id, run_id)
with open(save_path, 'wb') as f:
stats = [timestamps, perfs]
pickle.dump([validation_score, test_score, stats], f)
def test_cls():
save_dir = './data/eval_exps/soln-ml'
if not os.path.exists(save_dir):
os.makedirs(save_dir)
time_limit = 60
print('==> Start to evaluate with Budget %d' % time_limit)
ensemble_method = 'ensemble_selection'
eval_type = 'holdout'
iris = load_iris()
X, y = iris.data, iris.target
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.33,
random_state=1,
stratify=y)
dm = DataManager(X_train, y_train)
train_data = dm.get_data_node(X_train, y_train)
test_data = dm.get_data_node(X_test, y_test)
clf = Classifier(time_limit=time_limit,
output_dir=save_dir,
ensemble_method=ensemble_method,
enable_meta_algorithm_selection=False,
ensemble_size=10,
evaluation=eval_type,
metric='acc',
n_jobs=2)
clf.fit(train_data, tree_id=2)
print(clf.summary())
pred = clf.predict(test_data)
print(accuracy_score(test_data.data[1], pred))
shutil.rmtree(save_dir)
n_jobs = args.n_jobs
ensemble_method = args.ens_method
if ensemble_method == 'none':
ensemble_method = None
save_dir = './data/eval_exps/soln-ml'
if not os.path.exists(save_dir):
os.makedirs(save_dir)
print('==> Start to evaluate with Budget %d' % time_limit)
iris = load_iris()
X, y = iris.data, iris.target
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.33,
random_state=1)
dm = DataManager(X_train, y_train)
train_data = dm.get_data_node(X_train, y_train)
test_data = dm.get_data_node(X_test, y_test)
clf = Classifier(time_limit=time_limit,
output_dir=save_dir,
ensemble_method=ensemble_method,
evaluation=eval_type,
metric='acc',
n_jobs=n_jobs)
clf.fit(train_data)
pred = clf.predict(test_data)
print(balanced_accuracy_score(test_data.data[1], pred))
import numpy as np
import os
import sys
sys.path.append(os.getcwd())
from mindware.components.feature_engineering.transformations.preprocessor.text2vector import \
Text2VectorTransformation
from mindware.components.feature_engineering.transformation_graph import DataNode
from mindware.components.utils.constants import *
from mindware.estimators import Classifier
x = np.array([[1, 'I am good', 'I am right', 3],
[2, 'He is good', 'He is ok', 4],
[2.5, 'Everyone is good', 'Everyone is ok', 7],
[1.3333, 'well', 'what', 5]])
y = np.array([0, 1, 0, 1])
t2v = Text2VectorTransformation()
data = (x, y)
feature_type = [NUMERICAL, TEXT, TEXT, DISCRETE]
datanode = DataNode(data, feature_type)
clf = Classifier(time_limit=20,
enable_meta_algorithm_selection=False,
include_algorithms=['random_forest'])
clf.fit(datanode, opt_strategy='combined')
print(clf.predict(datanode))