def worker(d_indx, key, data):
print("Dataset: %s start" % key)
X, y = data
# b = GaussianNB()
# b = DecisionTreeClassifier(random_state=1410)
b = KNeighborsClassifier(weights='distance')
# b = MLPClassifier(random_state=1410)
# b = SVC(kernel='rbf', probability=False, random_state=1410)
n_estimators = 50
acc_prob = False
base_clf = StratifiedBagging(base_estimator=b,
ensemble_size=n_estimators,
acc_prob=True,
random_state=1410)
clfs = {
"MV-kNN":
StratifiedBagging(base_estimator=b,
ensemble_size=n_estimators,
acc_prob=False,
random_state=1410),
"ACC-kNN":
StratifiedBagging(base_estimator=b,
ensemble_size=n_estimators,
acc_prob=True,
random_state=1410),
"PRUNE-kNN":
CMOE(base_estimator=base_clf,
random_state=1410,
diversity="kw",
hard_voting=acc_prob)
}
n_splits = 5
n_repeats = 1
rskf = RepeatedStratifiedKFold(n_splits=n_splits,
n_repeats=n_repeats,
random_state=1410)
scores = np.zeros((32, n_splits * n_repeats, len(metrics)))
for fold_id, (train, test) in enumerate(rskf.split(X, y)):
print("FOLD %i: " % fold_id)
for clf_id, clf_name in enumerate(clfs):
clf = clone(clfs[clf_name])
clf = clfs[clf_name]
clf.fit(X[train], y[train])
y_pred = clf.predict(X[test])
if clf_id < 2:
for m_indx, (name, metric) in enumerate(metrics.items()):
scores[clf_id, fold_id, m_indx] = metric(y[test], y_pred)
else:
for e_n, ensemble_pred in enumerate(y_pred):
# print(ensemble_pred)
for m_indx, (name, metric) in enumerate(metrics.items()):
scores[clf_id + e_n, fold_id,
m_indx] = metric(y[test], ensemble_pred)
np.save("results/knn/%s_knn" % key, scores)
print("Dataset: %s end" % key)
def worker(d_indx, key, data):
print("Dataset: %s start" % key)
X, y = data
b = GaussianNB()
# b = DecisionTreeClassifier(random_state=1410)
# b = KNeighborsClassifier(weights='distance')
# b = MLPClassifier(random_state=1410)
# b = SVC(kernel='rbf', probability=False, random_state=1410)
n_estimators = 50
acc_prob = False
base_clf = StratifiedBagging(base_estimator = b, ensemble_size=n_estimators, acc_prob=True, random_state=1410)
clfs = {
"MV-GNB-ROS": StratifiedBagging(base_estimator = b, ensemble_size=n_estimators, acc_prob=True, random_state=1410, oversampled="ROS"),
"MV-GNB-SMOTE": StratifiedBagging(base_estimator = b, ensemble_size=n_estimators, acc_prob=True, random_state=1410, oversampled="SMOTE"),
"MV-GNB-SVMSMOTE": StratifiedBagging(base_estimator = b, ensemble_size=n_estimators, acc_prob=True, random_state=1410, oversampled="SVMSMOTE"),
"MV-GNB-B2SMOTE": StratifiedBagging(base_estimator = b, ensemble_size=n_estimators, acc_prob=True, random_state=1410, oversampled="B2SMOTE"),
}
n_splits = 5
n_repeats = 1
rskf = RepeatedStratifiedKFold(n_splits=n_splits, n_repeats=n_repeats, random_state=1410)
scores = np.zeros((len(clfs), n_splits * n_repeats, len(metrics)))
for fold_id, (train, test) in enumerate(rskf.split(X, y)):
print("FOLD %i: " % fold_id)
for clf_id, clf_name in enumerate(clfs):
clf = clone(clfs[clf_name])
clf = clfs[clf_name]
clf.fit(X[train], y[train])
y_pred = clf.predict(X[test])
for m_indx, (name, metric) in enumerate(metrics.items()):
scores[clf_id, fold_id, m_indx] = metric(y[test], y_pred)
np.save("results/gnb/%s_gnb_preproc" % key, scores)
print("Dataset: %s end" % key)
from skmultiflow.trees import HoeffdingTree
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
sea = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42))
knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
des="KNORAU1")
knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
des="KNORAU2")
knorae1 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
des="KNORAE1")
knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
des="KNORAE2")
clfs = (sea, knorau1, knorau2, knorae1, knorae2)
from sklearn.svm import SVC
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
none_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler="None"), oversampled="None", des="KNORAU1")
rus_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler="RUS"), oversampled="RUS", des="KNORAU1")
cnn_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAU1")
none_knorae1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler="None"), oversampled="None", des="KNORAE1")
rus_knorae1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler="RUS"), oversampled="RUS", des="KNORAE1")
cnn_knorae1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42), random_state=42, oversampler = "CNN"), oversampled="CNN" ,des="KNORAE1")
clfs = (none_knorau1, rus_knorau1, cnn_knorau1, none_knorae1, rus_knorae1, cnn_knorae1)
# Define worker
def worker(i, stream_n):
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.streams(random_state)
print(len(streams))
ob = OnlineBagging(n_estimators=20, base_estimator=GaussianNB())
oob = OOB(n_estimators=20, base_estimator=GaussianNB())
uob = UOB(n_estimators=20, base_estimator=GaussianNB())
ros_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAU2")
cnn_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
oversampler="CNN"),
oversampled="CNN",
des="KNORAU2")
ros_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAE2")
cnn_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
sea = SEA(base_estimator=StratifiedBagging(base_estimator=HoeffdingTree(
split_criterion='hellinger'),
random_state=42))
knorau1 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
des="KNORAU1")
knorau2 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
des="KNORAU2")
knorae1 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
des="KNORAE1")
knorae2 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
none_knorau1 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'),
random_state=42,
oversampler="None"),
oversampled="None",
des="KNORAU1")
ros_knorau1 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAU1")
b2_knorau1 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'),
random_state=42,
oversampler="B2"),
oversampled="B2",
des="KNORAU1")
)
import sys
from sklearn.base import clone
from sklearn.tree import DecisionTreeClassifier
from skmultiflow.trees import HoeffdingTree
# Select streams and methods
streams = h.realstreams()
print(len(streams))
ob = OnlineBagging(n_estimators=20, base_estimator=GaussianNB())
oob = OOB(n_estimators=20, base_estimator=GaussianNB())
uob = UOB(n_estimators=20, base_estimator=GaussianNB())
# sea = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42))
ros_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2")
cnn_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAU2")
ros_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAE2")
cnn_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(), random_state=42, oversampler = "CNN"), oversampled="CNN" ,des="KNORAE2")
clfs = (ob, oob, uob, ros_knorau2, cnn_knorau2, ros_knorae2, cnn_knorae2)
# Define worker
def worker(i, stream_n):
stream = streams[stream_n]
key = list(streams.keys())[i]
cclfs = [clone(clf) for clf in clfs]
from csm import SEA, StratifiedBagging, REA, LearnppCDS, LearnppNIE, OUSE, KMeanClustering, rea, TestThenTrain
from sklearn.naive_bayes import GaussianNB
from strlearn.streams import StreamGenerator
# from strlearn.evaluators import TestThenTrain
from strlearn.metrics import (balanced_accuracy_score, f1_score,
geometric_mean_score_1, precision, recall,
specificity)
import matplotlib.pyplot as plt
from scipy.ndimage.filters import gaussian_filter1d
import numpy as np
from sklearn.metrics import roc_auc_score
rea = REA(base_classifier=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
number_of_classifiers=5)
cds = LearnppCDS(base_classifier=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
number_of_classifiers=5)
nie = LearnppNIE(base_classifier=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
number_of_classifiers=5)
ouse = OUSE(base_classifier=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
number_of_classifiers=5)
kmc = KMeanClustering(base_classifier=StratifiedBagging(
base_estimator=GaussianNB(), random_state=42),
number_of_classifiers=5)
sea = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42),
from skmultiflow.trees import HoeffdingTree
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
none_knorau2 = SEA(base_estimator=StratifiedBagging(
base_estimator=GaussianNB(), random_state=42, oversampler="None"),
oversampled="None",
des="KNORAU2")
ros_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAU2")
b2_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(),
random_state=42,
oversampler="B2"),
oversampled="B2",
des="KNORAU2")
none_knorae2 = SEA(base_estimator=StratifiedBagging(
base_estimator=GaussianNB(), random_state=42, oversampler="None"),
oversampled="None",
recall,
specificity
)
import sys
from sklearn.base import clone
from sklearn.tree import DecisionTreeClassifier
from skmultiflow.trees import HoeffdingTree
import time
import matplotlib.pyplot as plt
# Select streams and methods
streams = h.timestream(100)
print(len(streams))
ros_knorau2_3 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2", n_estimators=3)
ros_knorau2_5 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAU2", n_estimators=5)
ros_knorau2_10 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2", n_estimators=10)
ros_knorau2_15 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2", n_estimators=15)
ros_knorau2_30 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2", n_estimators=30)
# cnn_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
# ), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAU2")
# ros_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
# ), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAE2")
# cnn_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
# ), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAE2")
from sklearn.neighbors import KNeighborsClassifier
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
sea = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'), random_state=42))
knorau1 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'), random_state=42),
des="KNORAU1")
knorau2 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'), random_state=42),
des="KNORAU2")
knorae1 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'), random_state=42),
des="KNORAE1")
knorae2 = SEA(base_estimator=StratifiedBagging(
base_estimator=KNeighborsClassifier(weights='distance'), random_state=42),
des="KNORAE2")
clfs = (sea, knorau1, knorau2, knorae1, knorae2)
# Define worker
from sklearn.svm import SVC
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
sea = SEA(base_estimator=StratifiedBagging(
base_estimator=SVC(probability=True, random_state=42), random_state=42))
knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42),
des="KNORAU1")
knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42),
des="KNORAU2")
knorae1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42),
des="KNORAE1")
knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42),
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
none_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42,
oversampler="None"),
oversampled="None",
des="KNORAU1")
ros_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAU1")
b2_knorau1 = SEA(base_estimator=StratifiedBagging(base_estimator=SVC(
probability=True, random_state=42),
random_state=42,
oversampler="B2"),
oversampled="B2",
des="KNORAU1")
geometric_mean_score_1,
precision,
recall,
specificity
)
import sys
from sklearn.base import clone
from sklearn.tree import DecisionTreeClassifier
from skmultiflow.trees import HoeffdingTree
# Select streams and methods
streams = h.moa_streams()
print(len(streams))
rea = REA(base_classifier=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42), number_of_classifiers=5)
cds = LearnppCDS(base_classifier=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42), number_of_classifiers=5)
nie = LearnppNIE(base_classifier=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42), number_of_classifiers=5)
ouse = OUSE(base_classifier=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42), number_of_classifiers=5)
kmc = KMeanClustering(base_classifier=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42), number_of_classifiers=5)
ros_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAU2")
cnn_knorau2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="CNN"), oversampled="CNN", des="KNORAU2")
ros_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(
), random_state=42, oversampler="ROS"), oversampled="ROS", des="KNORAE2")
cnn_knorae2 = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(), random_state=42, oversampler = "CNN"), oversampled="CNN" ,des="KNORAE2")
from sklearn.neural_network import MLPClassifier
if len(sys.argv) != 2:
print("PODAJ RS")
exit()
else:
random_state = int(sys.argv[1])
print(random_state)
# Select streams and methods
streams = h.toystreams(random_state)
print(len(streams))
gnb = SEA(base_estimator=StratifiedBagging(base_estimator=GaussianNB(), random_state=42))
ht = SEA(base_estimator=StratifiedBagging(base_estimator=HoeffdingTree(), random_state=42))
clfs = (gnb, ht)
# Define worker
def worker(i, stream_n):
stream = streams[stream_n]
cclfs = [clone(clf) for clf in clfs]
print("Starting stream %i/%i" % (i + 1, len(streams)))
eval = TestThenTrain(metrics=(
balanced_accuracy_score,
geometric_mean_score_1,
f1_score,
from sklearn.base import clone
from sklearn.tree import DecisionTreeClassifier
from skmultiflow.trees import HoeffdingTree
# Select streams and methods
streams = h.realstreams()
print(len(streams))
ob = OnlineBagging(n_estimators=20,
base_estimator=HoeffdingTree(split_criterion='hellinger'))
oob = OOB(n_estimators=20,
base_estimator=HoeffdingTree(split_criterion='hellinger'))
uob = UOB(n_estimators=20,
base_estimator=HoeffdingTree(split_criterion='hellinger'))
ros_knorau2 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAU2")
cnn_knorau2 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42,
oversampler="CNN"),
oversampled="CNN",
des="KNORAU2")
ros_knorae2 = SEA(base_estimator=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42,
oversampler="ROS"),
oversampled="ROS",
des="KNORAE2")
from strlearn.evaluators import TestThenTrain
from sklearn.naive_bayes import GaussianNB
from strlearn.metrics import (balanced_accuracy_score, f1_score,
geometric_mean_score_1, precision, recall,
specificity)
import sys
from sklearn.base import clone
from sklearn.tree import DecisionTreeClassifier
from skmultiflow.trees import HoeffdingTree
# Select streams and methods
streams = h.realstreams2()
print(len(streams))
rea = REA(base_classifier=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
number_of_classifiers=5)
cds = LearnppCDS(base_classifier=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
number_of_classifiers=5)
nie = LearnppNIE(base_classifier=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
number_of_classifiers=5)
ouse = OUSE(base_classifier=StratifiedBagging(
base_estimator=HoeffdingTree(split_criterion='hellinger'),
random_state=42),
number_of_classifiers=5)
kmc = KMeanClustering(base_classifier=StratifiedBagging(