def test_fit_biginc(self):
model = hyperopt_estimator(
classifier=components.any_classifier('classifier'),
verbose=1, max_evals=5, trial_timeout=5.0, fit_increment=20)
model.fit(self.X, self.Y)
# -- make sure we only get 5 even with big fit_increment
assert len(model.trials.trials) == 5
def test_sparse_random_projection(self):
# restrict n_components to be less than or equal to data dimension
# to prevent sklearn warnings from printing during tests
n_components = scope.int(
hp.quniform('preprocessing.n_components', low=1, high=8, q=1))
model = hyperopt_estimator(
classifier=components.gaussian_nb('classifier'),
preprocessing=[
components.sparse_random_projection(
'preprocessing',
n_components=n_components,
)
],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
X_train = np.random.randn(1000, 8)
Y_train = (self.X_train[:, 0] > 0).astype('int')
X_test = np.random.randn(1000, 8)
Y_test = (self.X_test[:, 0] > 0).astype('int')
model.fit(X_train, Y_train)
model.score(X_test, Y_test)
def test_fit_iter_basic(self):
model = hyperopt_estimator(verbose=1, trial_timeout=5.0)
for ii, trials in enumerate(model.fit_iter(self.X, self.Y)):
assert trials is model.trials
assert len(trials.trials) == ii
if ii == 10:
break
def sklearn_digits( classifier, algorithm, max_evals=100, seed=1,
filename = 'none', preproc=[], loss=None ):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename+'.dump'
estim = hyperopt_estimator( classifier=classifier, algo=algorithm,
preprocessing=preproc,
max_evals=max_evals, trial_timeout=60,
fit_increment_dump_filename=dump_file,
loss_fn=loss, verbose=1)
filename = filename + '.out'
digits = load_digits()
X = digits.data
y = digits.target
test_size = int( 0.2 * len( y ) )
np.random.seed( seed )
indices = np.random.permutation(len(X))
X_train = X[ indices[:-test_size]]
y_train = y[ indices[:-test_size]]
X_test = X[ indices[-test_size:]]
y_test = y[ indices[-test_size:]]
print(y_train.shape)
print(y_test.shape)
find_model( X_train, y_train, X_test, y_test, estim, filename )
def test_fit_iter_basic(self):
model = hyperopt_estimator(verbose=1, trial_timeout=5.0)
for ii, trials in enumerate(model.fit_iter(self.X, self.Y)):
assert trials is model.trials
assert len(trials.trials) == ii
if ii == 10:
break
def test_continuous_loss_fn():
"""
Demonstrate using a custom loss function with the continuous_loss_fn
option.
"""
from sklearn.metrics import log_loss
# Generate some random data
X = np.hstack([
np.vstack([
np.random.normal(0,1,size=(1000,10)),
np.random.normal(1,1,size=(1000,10)),
]),
np.random.normal(0,1,size=(2000,10)),
])
y = np.zeros(2000)
y[:1000] = 1
def loss_function(targ, pred):
# hyperopt_estimator flattens the prediction when saving it. This also
# affects multilabel classification.
pred = pred.reshape( (-1, 2) )
return log_loss(targ, pred[:,1])
# Try to fit an SGD model using log_loss as the loss function
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
loss_fn = loss_function,
continuous_loss_fn=True,
)
cls.fit(X,y,cv_shuffle=True)
def train_one_model(
name_classifier,
X: np.array,
y: np.array,
mix_algo,
max_evals: int,
timeout: int,
n: int,
) -> Tuple:
name, classifier = name_classifier
trainlogger.info("i'm using a timeout of {}".format(timeout))
m = hyperopt_estimator(
classifier=classifier("classifier"),
algo=mix_algo,
trial_timeout=timeout,
preprocessing=[],
max_evals=max_evals,
loss_fn=
f1lossfn, # f1 macro is probably more meaningfull than accuracy
# continuous_loss_fn = True,
seed=RANDOM_SEED,
)
trainlogger.info("training {}".format(name))
m.fit(
X, y, cv_shuffle=True,
n_folds=n) # hyperopt-sklearn takes care of the cross validations
m.retrain_best_model_on_full_data(X, y)
m = m.best_model()["learner"]
return (name, m)
def sklearn_digits( classifier=None ):
#estim = hyperopt_estimator( classifier=any_classifier('hai'), algo=tpe.suggest )
if classifier is None:
classifier = any_classifier('any')
estim = hyperopt_estimator( classifier=classifier )
digits = load_digits()
X = digits.data
y = digits.target
test_size = 50
np.random.seed(0)
indices = np.random.permutation(len(X))
X_train = X[ indices[:-test_size]]
y_train = y[ indices[:-test_size]]
X_test = X[ indices[-test_size:]]
y_test = y[ indices[-test_size:]]
estim.fit( X_train, y_train )
pred = estim.predict( X_test )
print( pred )
print ( y_test )
print( score( pred, y_test ) )
print( estim.best_model() )
def test_sparse_input():
"""
Ensure the estimator can handle sparse X matrices.
"""
import scipy.sparse as ss
# Generate some random sparse data
nrows,ncols,nnz = 100,50,10
ntrue = nrows // 2
D,C,R = [],[],[]
for r in range(nrows):
feats = np.random.choice(range(ncols), size=nnz, replace=False)
D.extend([1]*nnz)
C.extend(feats)
R.extend([r]*nnz)
X = ss.csr_matrix( (D,(R,C)), shape=(nrows, ncols))
y = np.zeros( nrows )
y[:ntrue] = 1
# Try to fit an SGD model
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
)
cls.fit(X,y)
def test_sparse_input():
"""
Ensure the estimator can handle sparse X matrices.
"""
import scipy.sparse as ss
# Generate some random sparse data
nrows, ncols, nnz = 100, 50, 10
ntrue = nrows // 2
D, C, R = [], [], []
for r in range(nrows):
feats = np.random.choice(range(ncols), size=nnz, replace=False)
D.extend([1] * nnz)
C.extend(feats)
R.extend([r] * nnz)
X = ss.csr_matrix((D, (R, C)), shape=(nrows, ncols))
y = np.zeros(nrows)
y[:ntrue] = 1
# Try to fit an SGD model
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
)
cls.fit(X, y)
def test_sparse_random_projection(self):
# restrict n_components to be less than or equal to data dimension
# to prevent sklearn warnings from printing during tests
n_components = scope.int(hp.quniform(
'preprocessing.n_components', low=1, high=8, q=1
))
model = hyperopt_estimator(
classifier=components.gaussian_nb('classifier'),
preprocessing=[
components.sparse_random_projection(
'preprocessing',
n_components=n_components,
)
],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
X_train = np.random.randn(1000, 8)
Y_train = (self.X_train[:, 0] > 0).astype('int')
X_test = np.random.randn(1000, 8)
Y_test = (self.X_test[:, 0] > 0).astype('int')
model.fit(X_train, Y_train)
model.score(X_test, Y_test)
def test_continuous_loss_fn():
"""
Demonstrate using a custom loss function with the continuous_loss_fn
option.
"""
from sklearn.metrics import log_loss
# Generate some random data
X = np.hstack([
np.vstack([
np.random.normal(0, 1, size=(1000, 10)),
np.random.normal(1, 1, size=(1000, 10)),
]),
np.random.normal(0, 1, size=(2000, 10)),
])
y = np.zeros(2000)
y[:1000] = 1
def loss_function(targ, pred):
# hyperopt_estimator flattens the prediction when saving it. This also
# affects multilabel classification.
pred = pred.reshape((-1, 2))
return log_loss(targ, pred[:, 1])
# Try to fit an SGD model using log_loss as the loss function
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
loss_fn=loss_function,
continuous_loss_fn=True,
)
cls.fit(X, y, cv_shuffle=True)
def test_smoke(self):
# -- verify the space argument is accepted and runs
space = components.generic_space()
model = hyperopt_estimator(verbose=1,
max_evals=10,
trial_timeout=5,
space=space)
model.fit(self.X, self.Y)
def test_fit(self):
model = hyperopt_estimator(
classifier=components.any_classifier('classifier'),
verbose=1,
max_evals=5,
trial_timeout=5.0)
model.fit(self.X, self.Y)
assert len(model.trials.trials) == 5
def test_regressor(self):
model = hyperopt_estimator(regressor=reg_fn('regressor'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=50.0,
max_evals=2,
verbose=True)
model.fit(self.X_train, self.Y_train)
model.score(self.X_test, self.Y_test)
def hyperopt_850_556():
# Load data
dir_key = '1406'
data_key = '850+556'
dir_path = dir_path_dict[dir_key]
data_str = dir_path + data_str_dict[data_key]
# Redirect stdout to file
stdout_path = 'outcome_hyperopt_any.any.txt'
print '[INFO] stdout_path:\t{}'.format(stdout_path)
sys.stdout = open(stdout_path, 'w')
print "[INFO] params:\talgo=tpe.suggest"
# Train
scores = []
sensis = []
specis = []
for i in range(10):
# Load data
data_path = data_str.format(i + 1)
print data_path
trainset, testset = get_dataset(data_path=data_path)
train_data, train_label = trainset
test_data, test_label = testset
# Create the estimator object
# estim = hyperopt_estimator(classifier=any_classifier('mySVC'),
# algo=tpe.suggest,
# preprocessing=[standard_scaler('std_scl')])
estim = hyperopt_estimator(algo=tpe.suggest, seed=RANDOM_SEED)
# Search the space of classifiers and preprocessing steps and their
# respective hyperparameters in sklearn to fit a model to the data
estim.fit(train_data, train_label)
# show instances of the best classifier
model = estim.best_model()
print model
# Make a prediction using the optimized model
prediction = estim.predict(test_data)
error = np.count_nonzero(prediction - test_label) / test_data.shape[0]
sensi, speci = my_scores(test_label, prediction)
print 1 - error, sensi, speci
# Report the accuracy of the classifier on a given set of data
score = estim.score(test_data, test_label)
print score
scores.append(score)
sensis.append(sensi)
specis.append(speci)
print scores
print "accur:\t{}\tstd:\t{}".format(np.mean(scores), np.std(scores))
print "sensi:\t{}".format(np.mean(sensis))
print "speci:\t{}".format(np.mean(specis))
def test_classifier(self):
model = hyperopt_estimator(
classifier=clf_fn('classifier'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
model.fit(self.X_train, self.Y_train_multilabel)
model.score(self.X_test, self.Y_test_multilabel)
def test_classifier(self):
model = hyperopt_estimator(
classifier=clf_fn('classifier'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
model.fit(self.X_train, self.Y_train_multilabel)
model.score(self.X_test, self.Y_test_multilabel)
def test_regressor(self):
model = hyperopt_estimator(
regressor=reg_fn('regressor'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
model.fit(self.X_train, self.Y_train)
model.score(self.X_test, self.Y_test)
def test_preprocessing(self):
model = hyperopt_estimator(
classifier=components.gaussian_nb('classifier'),
preprocessing=[pre_fn('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
model.fit(self.X_train, self.Y_train)
model.score(self.X_test, self.Y_test)
def test_preprocessing(self):
model = hyperopt_estimator(
classifier=components.gaussian_nb('classifier'),
preprocessing=[pre_fn('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
model.fit(self.X_train, self.Y_train)
model.score(self.X_test, self.Y_test)
def test_multinomial_nb(self):
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
# Inputs for MultinomialNB must be non-negative
model.fit(np.abs(self.X_train), self.Y_train)
model.score(np.abs(self.X_test), self.Y_test)
def test_multinomial_nb(self):
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
# Inputs for MultinomialNB must be non-negative
model.fit(np.abs(self.X_train), self.Y_train)
model.score(np.abs(self.X_test), self.Y_test)
def sklearn_newsgroups(classifier,
algorithm,
max_evals=100,
seed=1,
filename='none',
preproc=[],
loss=None):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename + '.dump'
estim = hyperopt_estimator(classifier=classifier,
algo=algorithm,
preprocessing=[tfidf('tfidf')],
max_evals=max_evals,
trial_timeout=240,
fit_increment_dump_filename=dump_file,
loss_fn=loss)
filename = filename + '.out'
if REMOVE_HEADERS:
train = fetch_20newsgroups(subset='train',
remove=('headers', 'footers', 'quotes'))
test = fetch_20newsgroups(subset='test',
remove=('headers', 'footers', 'quotes'))
else:
train = fetch_20newsgroups(subset='train')
test = fetch_20newsgroups(subset='test')
if PRE_VECTORIZE:
vectorizer = TfidfVectorizer()
X_train = vectorizer.fit_transform(train.data)
y_train = train.target
X_test = vectorizer.transform(test.data)
y_test = test.target
else:
X_train = train.data
y_train = train.target
X_test = test.data
y_test = test.target
print(y_train.shape)
print(y_test.shape)
find_model(X_train, y_train, X_test, y_test, estim, filename)
def test_one_hot_encoder(self):
# requires a classifier that can handle sparse data
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[components.one_hot_encoder('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
# Inputs for one_hot_encoder must be non-negative integers
model.fit(np.abs(np.round(self.X_test).astype(np.int)), self.Y_test)
model.score(np.abs(np.round(self.X_test).astype(np.int)), self.Y_test)
def test_one_hot_encoder(self):
# requires a classifier that can handle sparse data
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[components.one_hot_encoder('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
# Inputs for one_hot_encoder must be non-negative integers
model.fit(np.abs(np.round(self.X_test).astype(np.int)), self.Y_test)
model.score(np.abs(np.round(self.X_test).astype(np.int)), self.Y_test)
def test_warm_start(self):
model = hyperopt_estimator(
classifier=components.any_classifier('classifier'),
verbose=1, max_evals=5, trial_timeout=5.0)
params = model.get_params()
assert params['algo'] == rand.suggest
assert params['max_evals'] == 5
model.fit(self.X, self.Y, warm_start=False)
assert len(model.trials.trials) == 5
model.set_params(algo=tpe.suggest, max_evals=10)
params = model.get_params()
assert params['algo'] == tpe.suggest
assert params['max_evals'] == 10
model.fit(self.X, self.Y, warm_start=True)
assert len(model.trials.trials) == 15 # 5 + 10 = 15.
def sklearn_newsgroups( classifier, algorithm, max_evals=100, seed=1,
filename='none', preproc=[], loss=None ):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename+'.dump'
estim = hyperopt_estimator( classifier=classifier, algo=algorithm,
preprocessing=[tfidf('tfidf')],
max_evals=max_evals, trial_timeout=240,
fit_increment_dump_filename=dump_file,
loss_fn=loss)
filename = filename + '.out'
if REMOVE_HEADERS:
train = fetch_20newsgroups( subset='train',
remove=('headers', 'footers', 'quotes') )
test = fetch_20newsgroups( subset='test',
remove=('headers', 'footers', 'quotes') )
else:
train = fetch_20newsgroups( subset='train' )
test = fetch_20newsgroups( subset='test' )
if PRE_VECTORIZE:
vectorizer = TfidfVectorizer()
X_train = vectorizer.fit_transform( train.data )
y_train = train.target
X_test = vectorizer.transform( test.data )
y_test = test.target
else:
X_train = train.data
y_train = train.target
X_test = test.data
y_test = test.target
print(y_train.shape)
print(y_test.shape)
find_model( X_train, y_train, X_test, y_test, estim, filename )
def sklearn_convex(classifier,
algorithm,
max_evals=100,
seed=1,
filename='none',
preproc=[],
loss=None):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename + '.dump'
estim = hyperopt_estimator(classifier=classifier,
algo=algorithm,
preprocessing=preproc,
max_evals=max_evals,
trial_timeout=240,
fit_increment_dump_filename=dump_file,
loss_fn=loss)
filename = filename + '.out'
dataset_store.download('convex')
trainset, validset, testset = dataset_store.get_classification_problem(
'convex')
X_train = trainset.data.mem_data[0]
y_train = trainset.data.mem_data[1]
X_valid = validset.data.mem_data[0]
y_valid = validset.data.mem_data[1]
X_test = testset.data.mem_data[0]
y_test = testset.data.mem_data[1]
X_fulltrain = np.concatenate((X_train, X_valid))
y_fulltrain = np.concatenate((y_train, y_valid))
print(y_train.shape)
print(y_valid.shape)
print(y_test.shape)
#find_model( X_train, y_train, X_test, y_test, estim, filename )
find_model(X_fulltrain, y_fulltrain, X_test, y_test, estim, filename)
def sklearn_digits(classifier,
algorithm,
max_evals=100,
seed=1,
filename='none',
preproc=[],
loss=None):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename + '.dump'
estim = hyperopt_estimator(classifier=classifier,
algo=algorithm,
preprocessing=preproc,
max_evals=max_evals,
trial_timeout=60,
fit_increment_dump_filename=dump_file,
loss_fn=loss,
verbose=1)
filename = filename + '.out'
digits = load_digits()
X = digits.data
y = digits.target
test_size = int(0.2 * len(y))
np.random.seed(seed)
indices = np.random.permutation(len(X))
X_train = X[indices[:-test_size]]
y_train = y[indices[:-test_size]]
X_test = X[indices[-test_size:]]
y_test = y[indices[-test_size:]]
print(y_train.shape)
print(y_test.shape)
find_model(X_train, y_train, X_test, y_test, estim, filename)
def test_tfidf(self):
# requires a classifier that can handle sparse data
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[components.tfidf('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
X = np.array([
'This is the first document.',
'This document is the second document.',
'And this is the third one.',
'Is this the first document?',
])
Y = np.array([0, 1, 2, 0])
model.fit(X, Y)
model.score(X, Y)
def test_crossvalidation():
"""
Demonstrate performing a k-fold CV using the fit() method.
"""
# Generate some random data
X = np.hstack([
np.vstack([
np.random.normal(0,1,size=(1000,10)),
np.random.normal(1,1,size=(1000,10)),
]),
np.random.normal(0,1,size=(2000,10)),
])
y = np.zeros(2000)
y[:1000] = 1
# Try to fit a model
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
)
cls.fit(X,y,cv_shuffle=True, n_folds=5)
def test_tfidf(self):
# requires a classifier that can handle sparse data
model = hyperopt_estimator(
classifier=components.multinomial_nb('classifier'),
preprocessing=[components.tfidf('preprocessing')],
algo=rand.suggest,
trial_timeout=5.0,
max_evals=5,
)
X = np.array([
'This is the first document.',
'This document is the second document.',
'And this is the third one.',
'Is this the first document?',
])
Y = np.array([0, 1, 2, 0])
model.fit(X, Y)
model.score(X, Y)
def test_crossvalidation():
"""
Demonstrate performing a k-fold CV using the fit() method.
"""
# Generate some random data
X = np.hstack([
np.vstack([
np.random.normal(0, 1, size=(1000, 10)),
np.random.normal(1, 1, size=(1000, 10)),
]),
np.random.normal(0, 1, size=(2000, 10)),
])
y = np.zeros(2000)
y[:1000] = 1
# Try to fit a model
cls = hyperopt_estimator(
classifier=components.sgd('sgd', loss='log'),
preprocessing=[],
)
cls.fit(X, y, cv_shuffle=True, n_folds=5)
def sklearn_convex( classifier, algorithm, max_evals=100, seed=1,
filename = 'none', preproc=[], loss=None ):
global suppress_output
if suppress_output:
dump_file = None
else:
dump_file = filename+'.dump'
estim = hyperopt_estimator( classifier=classifier, algo=algorithm,
preprocessing=preproc,
max_evals=max_evals, trial_timeout=240,
fit_increment_dump_filename=dump_file,
loss_fn=loss)
filename = filename + '.out'
dataset_store.download('convex')
trainset,validset,testset = dataset_store.get_classification_problem('convex')
X_train = trainset.data.mem_data[0]
y_train = trainset.data.mem_data[1]
X_valid = validset.data.mem_data[0]
y_valid = validset.data.mem_data[1]
X_test = testset.data.mem_data[0]
y_test = testset.data.mem_data[1]
X_fulltrain = np.concatenate((X_train, X_valid))
y_fulltrain = np.concatenate((y_train, y_valid))
print(y_train.shape)
print(y_valid.shape)
print(y_test.shape)
#find_model( X_train, y_train, X_test, y_test, estim, filename )
find_model( X_fulltrain, y_fulltrain, X_test, y_test, estim, filename )
def mnist_digits():
estim = hyperopt_estimator( classifier=any_classifier('hai') )
digits = fetch_mldata('MNIST original')
X = digits.data
y = digits.target
test_size = int( 0.2 * len( y ) )
np.random.seed(0)
indices = np.random.permutation(len(X))
X_train = X[ indices[:-test_size]]
y_train = y[ indices[:-test_size]]
X_test = X[ indices[-test_size:]]
y_test = y[ indices[-test_size:]]
estim.fit( X_train, y_train )
pred = estim.predict( X_test )
print( pred )
print ( y_test )
print( score( pred, y_test ) )
print( estim.best_model() )
def main():
experiment = Experiment(
api_key=os.getenv("COMET_API_KEY", None),
project_name="mof-oxidation-states",
)
print("Loading Data")
X_train = np.load(FEAT_TRAIN_PATH)
X_valid = np.load(FEAT_VALID_PATH)
X_test = np.load(FEAT_TEST_PATH)
scaler = StandardScaler()
X_train = scaler.fit_transform(X_train)
X_valid = scaler.transform(X_valid)
X_test = scaler.transform(X_test)
joblib.dump(
scaler, os.path.join("models",
STARTTIMESTRING + "_" + "scaler.joblib"))
# vt = VarianceThreshold(0.1)
# X_train = vt.fit_transform(X_train)
# X_valid = vt.transform(X_valid)
# X_test = vt.transform(X_test)
# joblib.dump(vt, os.path.join("models", STARTTIMESTRING + "_" + "vt.joblib"))
y_train = np.load(LABEL_TRAIN_PATH)
y_valid = np.load(LABEL_VALID_PATH)
y_test = np.load(LABEL_TEST_PATH)
optimized_models = []
mix_algo = partial(
mix.suggest,
p_suggest=[
(0.15, rand.suggest),
(0.7, tpe.suggest),
(0.15, anneal.suggest),
],
)
print("Optimizing classifiers")
for name, classifier in CLASSIFIERS:
m = hyperopt_estimator(
classifier=classifier("classifier"),
algo=mix_algo,
trial_timeout=TIMEOUT,
loss_fn=f1_loss,
preprocessing=[],
max_evals=MAX_EVALS,
seed=RANDOM_SEED,
)
# Random undersampling to make the base estimators even more uncorrelated
X_train_, y_train_ = undersample_2(X_train, y_train)
m.fit(
np.vstack([X_train_, X_valid]),
np.vstack([y_train_.reshape(-1, 1),
y_valid.reshape(-1, 1)]),
valid_size=len(X_valid),
cv_shuffle=False,
)
m.retrain_best_model_on_full_data(X_train_, y_train_)
m = m.best_model()["learner"]
optimized_models.append((name, m))
model_eval(
optimized_models,
X_train,
y_train,
X_test,
y_test,
outdir_metrics="metrics",
outdir_models="models",
)
vc = VotingClassifier(optimized_models, voting="soft")
vc._calibrate_base_estimators("sigmoid", X_valid, y_valid)
model_eval(
[("ensemble", vc)],
X_train,
y_train,
X_test,
y_test,
outdir_metrics="metrics",
outdir_models="models",
)
def test_fit(self):
model = hyperopt_estimator(
classifier=components.any_classifier('classifier'),
verbose=1, max_evals=5, trial_timeout=5.0)
model.fit(self.X, self.Y)
assert len(model.trials.trials) == 5
def test_fit(self):
model = hyperopt_estimator(verbose=1, max_evals=5, trial_timeout=5.0)
model.fit(self.X, self.Y)
assert len(model.trials.trials) == 5
def test_smoke(self):
# -- verify the space argument is accepted and runs
space = components.generic_space()
model = hyperopt_estimator(
verbose=1, max_evals=10, trial_timeout=5, space=space)
model.fit(self.X, self.Y)
def test_fit(self):
model = hyperopt_estimator(verbose=1, max_evals=5, trial_timeout=5.0)
model.fit(self.X, self.Y)
assert len(model.trials.trials) == 5
def tune_fit( # pylint:disable=dangerous-default-value
models: list,
X: np.ndarray,
y: np.ndarray,
max_evals: int = 400,
timeout: int = 10 * 60,
mix_ratios: dict = {"rand": 0.1, "tpe": 0.8, "anneal": 0.1},
valid_size: float = VALID_SIZE,
) -> list:
"""Tune model hyperparameters using hyperopt using a mixed strategy.
Make sure when using this function that no data leakage happens.
This data here should be seperate from training and test set.
Arguments:
models {list} -- list of models that should be optimized
X_valid {np.ndarray} -- features
y_valid {np.ndarray} -- labels
max_evals {int} -- maximum number of evaluations of hyperparameter optimizations
timeout {int} -- timeout in seconds after which the optimization stops
mix_ratios {dict} -- dictionary which provides the ratios of the different optimization algorithms
valid_size {float} -- fraction of the last part of the training set used for validation
Returns:
list -- list of tuples (name, model) of optimized models
"""
assert sum(list(mix_ratios.values())) == 1
assert list(mix_ratios.keys()) == ["rand", "tpe", "anneal"]
trainlogger.debug("performing hyperparameter optimization")
optimized_models = []
mix_algo = partial(
mix.suggest,
p_suggest=[
(mix_ratios["rand"], rand.suggest),
(mix_ratios["tpe"], tpe.suggest),
(mix_ratios["anneal"], anneal.suggest),
],
)
for name, classifier in models:
m = hyperopt_estimator(
classifier=classifier("classifier"),
algo=mix_algo,
trial_timeout=timeout,
preprocessing=[],
max_evals=max_evals,
seed=RANDOM_SEED,
# n_jobs=-1, # todo fix installation to use my forks
)
m.fit(
X, y, valid_size=valid_size, cv_shuffle=False
) # avoid shuffleing to have the same validation set for the ensemble stage
# chose the model with best hyperparameters and train it
n_train = int(len(y) * (1 - valid_size))
X_train = X[:n_train]
y_train = y[:n_train]
m.retrain_best_model_on_full_data(X_train, y_train)
m = m.best_model()["learner"]
optimized_models.append((name, m))
return optimized_models
