def perform_over(**kwargs):
o2pf_obj = kwargs['o2pf_obj']
X = kwargs['X']
y = kwargs['y']
X_valid = kwargs['X_valid']
y_valid = kwargs['y_valid']
X_test = kwargs['X_test']
y_test = kwargs['y_test']
ds = kwargs['ds']
f = kwargs['f']
k_max = kwargs['k_max']
common = COMMON()
approach = o2pf_obj.__class__.__name__
best_k = common.optimization(X, y, X_valid, y_valid, o2pf_obj, k_max, ds,
f, approach, 'Results')
start_time = time()
o2pf_obj.k_max = best_k
all_x, all_y = o2pf_obj.fit_resample(X, y)
end_time = time() - start_time
common.saveDataset(all_x, all_y, pathDataset, approach)
common.saveResults(all_x, all_y, X_test, y_test, ds, f, approach,
o2pf_obj.min_class_label, end_time, 'Results', best_k)
def fit_resample(self, X, y, valid=None):
us_object = self.variant()
o2pf_ = O2PF()
X_res, y_res = us_object.fit_resample(X, y, valid)
self.min_class_label = us_object.min_class_label
if valid is not None:
X_valid = valid[:, :-1]
y_valid = valid[:, -1].astype(np.int)
common = COMMON()
best_k = common.optimization(X, y, X_valid, y_valid, o2pf_,
self.k_max)
else:
best_k = self.k_max[0]
o2pf_.k_max = best_k
all_x, all_y = o2pf_.fit_resample(X_res, y_res)
return all_x, all_y
def perform_under(**kwargs):
# Apply the undersampling according to the opf-us variant represented by the opf_us_obj
opf_us_obj = kwargs['us_obj']
X = kwargs['X']
y = kwargs['y']
X_test = kwargs['X_test']
y_test = kwargs['y_test']
f = kwargs['fold']
ds = kwargs['ds']
valid = kwargs['valid']
start_time = time()
X_res, y_res = opf_us_obj.fit_resample(X, y, valid)
end_time = time() - start_time
approach = opf_us_obj.__class__.__name__
common = COMMON()
common.saveTimeOnly(ds, f, approach, end_time, 'Results')
def perform_under(**kwargs):
# Apply the hybrid approach according to the opf-us variant represented by the hybrid_obj
hybrid_obj = kwargs['hybrid_obj']
X = kwargs['X']
y = kwargs['y']
X_test = kwargs['X_test']
y_test = kwargs['y_test']
f = kwargs['fold']
ds = kwargs['ds']
valid = kwargs['valid']
start_time = time()
all_x, all_y = hybrid_obj.fit_resample(X, y)
end_time = time() - start_time
approach = hybrid_obj.__class__.__name__
common = COMMON()
# Save the results of the oversampling
common.saveTimeOnly(ds, f, approach, end_time, 'Results')
def perform_over(**kwargs):
o2pf_obj = kwargs['o2pf_obj']
X = kwargs['X']
y = kwargs['y']
X_valid = kwargs['X_valid']
y_valid = kwargs['y_valid']
X_test = kwargs['X_test']
y_test = kwargs['y_test']
ds = kwargs['ds']
f = kwargs['f']
k_max = kwargs['k_max']
common = COMMON()
approach = o2pf_obj.__class__.__name__
best_k = 5
start_time = time()
o2pf_obj.k_max = best_k
all_x, all_y = o2pf_obj.fit_resample(X, y)
end_time = time() - start_time
common.saveTimeOnly(ds, f, approach, end_time, 'Results')
import numpy as np
from oversampling.o2pf import O2PF
from common.common import COMMON
import sys
if len(sys.argv) <= 2:
print('Usage: %s <train dataset> <k_max>' % sys.argv[0])
raise SystemExit
train = np.loadtxt(sys.argv[1],delimiter=',', dtype=np.float32)
X = train[:,:-1]
y = train[:,-1].astype(int)
o2pf = O2PF(k_max=int(sys.argv[2]))
X_res, y_res = o2pf.fit_resample(X, y)
path = 'data'
common = COMMON()
common.saveDataset(X_res, y_res, path, o2pf.__class__.__name__)
import numpy as np
from hybrid.us1_o2pf import US1O2PF
from common.common import COMMON
import sys
if len(sys.argv) <= 2:
print('Usage: %s <train dataset> <k_max> <validation dataset (optional)>' %
sys.argv[0])
raise SystemExit
train = np.loadtxt(sys.argv[1], delimiter=',', dtype=np.float32)
valid = None
if len(sys.argv) >= 4:
valid = np.loadtxt(sys.argv[3], delimiter=',', dtype=np.float32)
concat = np.concatenate((train, valid))
X = concat[:, :-1]
y = concat[:, -1].astype(np.int)
else:
X = train[:, :-1]
y = train[:, -1].astype(int)
hybrid_obj = US1O2PF(k_max=[int(sys.argv[2])])
X_res, y_res = hybrid_obj.fit_resample(X, y, valid)
path = 'data'
common = COMMON()
common.saveDataset(X_res, y_res, path, hybrid_obj.__class__.__name__)