def optimizeRF(did,amount):
X,y = read_did(did)
X = add_copy_features(X,amount)
X_train, X_test, y_train, y_test = train_test_split(X, y)
clf = RandomForestClassifier()
params = {'max_features': range(1,len(X[0])), 'min_samples_split': range(2,20)}
randomRF = RandomizedSearchCV(clf, param_distributions=params,
n_iter=40,n_jobs = 3)
with stopwatch() as sw:
_=randomRF.fit(X_train, y_train)
duration = sw.duration
estimator = randomRF.best_estimator_
cv_score = list(cross_val_score(randomRF.best_estimator_,X,y,cv = 10))
cv_score.append(randomRF.score(X_test,y_test))
return estimator,cv_score,duration
def optimizeADA(did,amount):
X,y = read_did(did)
iters = 40
X = add_copy_features(X,amount)
X_train, X_test, y_train, y_test = train_test_split(X, y)
clf = AdaBoostClassifier()
learns = [random()*1.9 + 0.1 for i in range(iters*4)]
params = {'learning_rate': learns}
randomRF = RandomizedSearchCV(clf, param_distributions=params,
n_iter=iters,n_jobs = 3)
with stopwatch() as sw:
_=randomRF.fit(X_train, y_train)
duration = sw.duration
estimator = randomRF.best_estimator_
cv_score = list(cross_val_score(randomRF.best_estimator_,X,y,cv = 10))
cv_score.append(randomRF.score(X_test,y_test))
return estimator,cv_score,duration
def functionoidFeatures(clf,did,rand,amount):
listOutput = []
local = True
if (local) :
X,y = read_did(did)
else:
DataSetOML = oml.datasets.get_dataset(did)
X, y = DataSetOML.get_data(target=DataSetOML.default_target_attribute);
if len(X[0]) > 10:
feature_X = add_noise_features(X,len(X[0])//amount,rand)
else:
feature_X = add_noise_features(X,10//amount,rand)
clf1 = clf
clf2 = copy(clf)
listOutput.append(cross_val_score(clf1, X, y,cv=10,n_jobs = -1,scoring = 'f1_weighted'))
listOutput.append(cross_val_score(clf2, feature_X, y,cv=10,n_jobs = -1, scoring = 'f1_weighted'))
return listOutput
def functionoid_amount(clf,did, amount,adj):
local = True
listOutput = []
if (local) :
X,y = read_did(did)
else:
DataSetOML = oml.datasets.get_dataset(did)
X, y = DataSetOML.get_data(target=DataSetOML.default_target_attribute);
svm1 = clf
svm2 = copy(svm1)
svm3 = copy(svm1)
X_train, X_test, y_train, y_test = train_test_split(X, y,stratify=y)
noise_X_train,noise_y_train = random_test_set4(X_train,y_train,amount)
#len(noise_X_train)
#len(noise_y_train)
noise_X_test,noise_y_test = random_test_set4(X_test,y_test,amount)
_ = svm1.fit(noise_X_train, noise_y_train)
_ = svm2.fit(X_train, y_train)
y_predict_noise1 = svm1.predict(noise_X_test)
y_predict_noise2 = svm2.predict(noise_X_test)
y_predict = svm2.predict(X_test)
#print("Score of no noise in training data",svm1.score(noise_X_test,noise_y_test))
#print("Score of noise in training and test data",svm2.score(noise_X_test,noise_y_test))
# add noise then split up
noise_X,noise_y = random_test_set4(X,y,amount)
X_train3, X_test3, y_train3, y_test3 = train_test_split(noise_X, noise_y, stratify = noise_y)
_ = svm3.fit(X_train3,y_train3)
y_predict_noise3 = svm3.predict(X_test3)
#print("Score of adding noise before split",svm3.score(X_test3,y_test3))
#calculate noise influence and errors made
score_1 = 0
score_2 = 0
score_3 = 0
noise1 = 0
noise2 = 0
noise3 = 0
score = 1
pred_wrong = set()
pred_wrong3 = set()
for i in range(0,len(noise_X_test)):
if noise_X_test[i] == [0] *len(noise_X_test[i]):
pred_wrong.add(i)
for i in range(0,len(noise_y_test)):
if y_predict_noise2[i] == noise_y_test[i]:
score_2 = score_2 + 1
if i in pred_wrong:
noise2 = noise2+1
if y_predict_noise1[i] == noise_y_test[i]:
score_1 = score_1 + 1
if i in pred_wrong:
noise1 = noise1+1
listOutput.append({'score' : score_1, 'noise' : noise1})
listOutput.append({'score' : score_2, 'noise' : noise2})
#print("amount of wrong classification first classifier ",((len(y_test)-score_1)/len(y_test)))
#print("amount of wrong classification second classifier ",((len(y_test)-score_2)/len(y_test)))
#print("noise classified by first ",noise1)
#print("noise classified by seconds", noise2)
for i in range(0,len(X_test3)):
if X_test3[i] == [0] *len(X_test3[i]):
pred_wrong3.add(i)
for i in range(0,len(y_test3)):
if y_predict_noise3[i] == y_test3[i]:
score_3 = score_3 + 1
if i in pred_wrong3:
noise3 = noise3+1
listOutput.append({'score' : score_3, 'noise' : noise3})
listOutput.append({'test1' : len(noise_y_test), 'test2' : len(y_test3),
'amountOfTargets' : len(values_target(y)), 'noise1' : len(pred_wrong),
'noise2' : len(pred_wrong3)})
#print("amount of wrong classification third classifier", ((len(y_test3)-score_3)/len(y_test3)))
#print("noise classified by third " ,noise3)
return listOutput
def loopFunctionoid_amount(clf,did,times,adj,amount):
listOutput = functionoid_amount(clf,did,amount,adj)
if (local) :
X,y = read_did(did)
else:
DataSetOML = oml.datasets.get_dataset(did)
X, y = DataSetOML.get_data(target=DataSetOML.default_target_attribute);
for i in range(1,times):
svm1 = copy(clf)
svm2 = copy(clf)
svm3 = copy(clf)
X_train, X_test, y_train, y_test = train_test_split(X, y,stratify=y)
noise_X_train,noise_y_train = add_noise_amount(X_train,y_train,amount,adj)
#len(noise_X_train)
#len(noise_y_train)
noise_X_test,noise_y_test = add_noise_amount(X_test,y_test,amount,adj)
_ = svm1.fit(noise_X_train,noise_y_train)
_ = svm2.fit(X_train,y_train)
y_predict_noise1 = svm1.predict(noise_X_test)
y_predict_noise2 = svm2.predict(noise_X_test)
y_predict = svm2.predict(X_test)
#print("Score of no noise in training data",svm1.score(noise_X_test,noise_y_test))
#print("Score of noise in training and test data",svm2.score(noise_X_test,noise_y_test))
# add noise then split up
noise_X,noise_y = add_noise_amount(X,y,amount,adj)
X_train3, X_test3, y_train3, y_test3 = train_test_split(noise_X, noise_y, stratify = noise_y)
_ = svm3.fit(X_train3,y_train3)
y_predict_noise3 = svm3.predict(X_test3)
#print("Score of adding noise before split",svm3.score(X_test3,y_test3))
#calculate noise influence and errors made
score_1 = 0
score_2 = 0
score_3 = 0
noise1 = 0
noise2 = 0
noise3 = 0
pred_wrong = set()
pred_wrong3 = set()
for i in range(0,len(noise_X_test)):
if noise_X_test[i] == [0] *len(noise_X_test[i]):
pred_wrong.add(i)
for i in range(0,len(noise_y_test)):
if y_predict_noise2[i] == noise_y_test[i]:
score_2 = score_2 + 1
if i in pred_wrong:
noise2 = noise2+1
if y_predict_noise1[i] == noise_y_test[i]:
score_1 = score_1 + 1
if i in pred_wrong:
noise1 = noise1+1
listOutput[0]['score'] = listOutput[0]['score'] + score_1
listOutput[1]['score'] = listOutput[1]['score'] + score_2
listOutput[0]['noise'] = listOutput[0]['noise'] + noise1
listOutput[1]['noise'] = listOutput[1]['noise'] + noise2
for i in range(0,len(X_test3)):
if X_test3[i] == [0] *len(X_test3[i]):
pred_wrong3.add(i)
for i in range(0,len(y_test3)):
if y_predict_noise3[i] == y_test3[i]:
score_3 = score_3 + 1
if i in pred_wrong3:
noise3 = noise3+1
listOutput[2]['score'] = listOutput[2]['score'] + score_3
listOutput[2]['noise'] = listOutput[2]['noise'] + noise3
listOutput[0]['score'] = listOutput[0]['score']//times
listOutput[1]['score'] = listOutput[1]['score']//times
listOutput[0]['noise'] = listOutput[0]['noise']//times
listOutput[1]['noise'] = listOutput[1]['noise']//times
listOutput[2]['score'] = listOutput[2]['score']//times
listOutput[2]['noise'] = listOutput[2]['noise']//times
return listOutput