def ADA_Classifier(X_train, X_cv, X_test, Y_train,Y_cv,Y_test, Actual_DS):
print("***************Starting AdaBoost Classifier***************")
t0 = time()
clf = AdaBoostClassifier(n_estimators=300)
clf.fit(X_train, Y_train)
preds = clf.predict(X_cv)
score = clf.score(X_cv,Y_cv)
print("AdaBoost Classifier - {0:.2f}%".format(100 * score))
Summary = pd.crosstab(label_enc.inverse_transform(Y_cv), label_enc.inverse_transform(preds),
rownames=['actual'], colnames=['preds'])
Summary['pct'] = (Summary.divide(Summary.sum(axis=1), axis=1)).max(axis=1)*100
print(Summary)
#Check with log loss function
epsilon = 1e-15
#ll_output = log_loss_func(Y_cv, preds, epsilon)
preds2 = clf.predict_proba(X_cv)
ll_output2= log_loss(Y_cv, preds2, eps=1e-15, normalize=True)
print(ll_output2)
print("done in %0.3fs" % (time() - t0))
preds3 = clf.predict_proba(X_test)
#preds4 = clf.predict_proba((Actual_DS.ix[:,'feat_1':]))
preds4 = clf.predict_proba(Actual_DS)
print("***************Ending AdaBoost Classifier***************")
return pd.DataFrame(preds2) , pd.DataFrame(preds3),pd.DataFrame(preds4)
def ab(train_data,train_label,val_data,val_label,test_data,name="adaboost_submission.csv"): print "Start training AdaBoost..." abClf = AdaBoostClassifier() abClf.fit(train_data,train_label) #evaluate on validation set val_pred_label = abClf.predict_proba(val_data) logloss = preprocess.evaluation(val_label,val_pred_label) print "logloss of validation set:",logloss print "Start classify test set..." test_label = abClf.predict_proba(test_data) preprocess.saveResult(test_label,filename = name)
def ab_predictedValue():
print '----------AdaBoost----------'
ab_clf = AdaBoostClassifier(n_estimators = NoOfEstimators)
ab_clf.fit(train_df[features], train_df['SeriousDlqin2yrs'])
ab_predictedValue = ab_clf.predict_proba(test_df[features])
print 'Feature Importance = %s' % ab_clf.feature_importances_
return ab_predictedValue[:,1]
def do_all_study(X,y):
names = [ "Decision Tree","Gradient Boosting",
"Random Forest", "AdaBoost", "Naive Bayes"]
classifiers = [
#SVC(),
DecisionTreeClassifier(max_depth=10),
GradientBoostingClassifier(max_depth=10, n_estimators=20, max_features=1),
RandomForestClassifier(max_depth=10, n_estimators=20, max_features=1),
AdaBoostClassifier()]
for name, clf in zip(names, classifiers):
estimator,score = plot_learning_curve(clf, X_train, y_train, scoring='roc_auc')
clf_GBC = GradientBoostingClassifier(max_depth=10, n_estimators=20, max_features=1)
param_name = 'n_estimators'
param_range = [1, 5, 10, 20,40]
plot_validation_curve(clf_GBC, X_train, y_train,
param_name, param_range, scoring='roc_auc')
clf_GBC.fit(X_train,y_train)
y_pred_GBC = clf_GBC.predict_proba(X_test)[:,1]
print("ROC AUC GradientBoostingClassifier: %0.4f" % roc_auc_score(y_test, y_pred_GBC))
clf_AB = AdaBoostClassifier()
param_name = 'n_estimators'
param_range = [1, 5, 10, 20,40]
plot_validation_curve(clf_AB, X_train, y_train,
param_name, param_range, scoring='roc_auc')
clf_AB.fit(X_train,y_train)
y_pred_AB = clf_AB.predict_proba(X_test)[:,1]
print("ROC AUC AdaBoost: %0.4f" % roc_auc_score(y_test, y_pred_AB))
def training(baseclassparameters, adaparameters, queue):
treeclassifier = DecisionTreeClassifier(**baseclassparameters)
adaclassifier = AdaBoostClassifier(treeclassifier, **adaparameters)
print "\nBegin calculation with {0} and {1}".format(str(baseclassparameters), str(adaparameters))
adaclassifier.fit(Xtrain, ytrain)
#Predict with the model
prob_predict_test = adaclassifier.predict_proba(Xtest)[:,1]
#Calculate maximal significance
True_Signal_test = prob_predict_test[ytest==1]
True_Bkg_test = prob_predict_test[ytest==0]
best_significance = 0
for x in np.linspace(0, 1, 1000):
S = float(len(True_Signal_test[True_Signal_test>x]))
B = float(len(True_Bkg_test[True_Bkg_test>x]))
significance = S/np.sqrt(S+B)
if significance > best_significance:
best_significance = significance
best_x = x
best_S = S
best_B = B
print "\nCalculation with {} and {} done ".format(str(baseclassparameters), str(adaparameters))
print "Best significance of {0:.2f} archived when cutting at {1:.3f}".format(best_significance, best_x)
print "Signal efficiency: {0:.2f}%".format(100.*best_S/len(True_Signal_test))
print "Background efficiency: {0:.2f}%".format(100.*best_B/len(True_Bkg_test))
print "Purity: {0:.2f}%".format(100.*best_S/(best_S+best_B))
queue.put( (best_significance, baseclassparameters, adaparameters) )
def test_oneclass_adaboost_proba():
# Test predict_proba robustness for one class label input.
# In response to issue #7501
# https://github.com/scikit-learn/scikit-learn/issues/7501
y_t = np.ones(len(X))
clf = AdaBoostClassifier().fit(X, y_t)
assert_array_almost_equal(clf.predict_proba(X), np.ones((len(X), 1)))
def ada_prediction(features_train, labels_train, features_test, ids):
X_train, X_test, y_train, y_test = cross_validation.train_test_split(features_train, labels_train, random_state=1301, stratify=labels_train, test_size=0.3)
clf = AdaBoostClassifier(RandomForestClassifier(bootstrap=True,
criterion='entropy', max_depth=None, max_features=2,
max_leaf_nodes=16, min_samples_split=10, n_estimators=1000,
n_jobs=-1, oob_score=False),
algorithm="SAMME",
n_estimators=200)
#clf_acc = clf.fit(X_train, y_train)
# print(clf.best_estimator_)
#feature_importance = clf.feature_importances_
#print (feature_importance)
#pred = clf_acc.predict_proba(X_test)[:,1]
#print (y_test, pred)
# acc = accuracy_score(y_test, pred)
# print ("Acc {}".format(acc))
clf = clf.fit(features_train, labels_train)
pred = clf.predict_proba(features_test)[:,1]
predictions_file = open("data/canivel_ada_forest.csv", "wb")
predictions_file_object = csv.writer(predictions_file)
predictions_file_object.writerow(["ID", "TARGET"])
predictions_file_object.writerows(zip(ids, pred))
predictions_file.close()
def ada_boost_cv(x_train,
y_train,
cv,
max_tree_depth,
n_estimators,
learning_rate):
tree_classifier = DecisionTreeClassifier(max_depth=max_tree_depth,
class_weight="balanced")
ada_boost_classifier = AdaBoostClassifier(base_estimator=tree_classifier,
n_estimators=n_estimators,
learning_rate=learning_rate)
y_bar = cross_val_predict(estimator=ada_boost_classifier,
X=x_train,
y=y_train,
cv=cv,
n_jobs=cv)
y_bar_proba = ada_boost_classifier.predict_proba(x_train)
print(list(zip(y_bar,y_bar_proba)))
cm = confusion_matrix(y_train,y_bar)
accuracy_negative = cm[0,0] / np.sum(cm[0,:])
accuracy_positive = cm[1,1] / np.sum(cm[1,:])
precision = cm[1,1] / (cm[1,1] + cm[0,1])
recall = cm[1,1] / (cm[1,1] + cm[1,0])
f1_score = 2 * precision * recall / (precision + recall)
return accuracy_positive, accuracy_negative, precision, recall, f1_score
def Adaboost(TrainData,TestData):
features=['Time','Season','Hour','Minute','District']
clf = AdaBoostClassifier(tree.DecisionTreeClassifier(),n_estimators=30)
size=[0.1,0.2,0.3,0.4,0.5,0.6,0.7,0.8,0.9]
for i in range(0,len(size)):
train,validation= train_test_split(TrainData, train_size=size[i])
while len(set(train['Category'])) != len(set(validation['Category'])):
train,validation= train_test_split(TrainData, train_size=size[i])
clf = clf.fit(train[features], train['Category'])
"""stop = timeit.default_timer()
print "Runnin time adaboost is ", stop-start"""
predicted=np.array(clf.predict_proba(validation[features]))
model=clf.predict(train[features])
model1=clf.predict(validation[features])
#scores = cross_val_score(clf, validation[features], validation['Category'])
#print "Scores mean is",scores.mean()
#accuracy
print "Training accuracy is", accuracy_score(train['Category'].values.tolist(),model)
print "Validation accuracy is",accuracy_score(validation['Category'].values.tolist(),model1)
print "Precision is ",precision_score(validation['Category'].values.tolist(),model1,average='macro')
print "Recall is ",recall_score(validation['Category'].values.tolist(),model1,average='macro')
print "Log loss is", log_loss(validation['Category'].values.tolist(),predicted,eps=1e-15, normalize=True, sample_weight=None)
#writing to file
"""Category_new=[]
def test_staged_predict():
"""Check staged predictions."""
# AdaBoost classification
for alg in ['SAMME', 'SAMME.R']:
clf = AdaBoostClassifier(algorithm=alg, n_estimators=10)
clf.fit(iris.data, iris.target)
predictions = clf.predict(iris.data)
staged_predictions = [p for p in clf.staged_predict(iris.data)]
proba = clf.predict_proba(iris.data)
staged_probas = [p for p in clf.staged_predict_proba(iris.data)]
score = clf.score(iris.data, iris.target)
staged_scores = [s for s in clf.staged_score(iris.data, iris.target)]
assert_equal(len(staged_predictions), 10)
assert_array_almost_equal(predictions, staged_predictions[-1])
assert_equal(len(staged_probas), 10)
assert_array_almost_equal(proba, staged_probas[-1])
assert_equal(len(staged_scores), 10)
assert_array_almost_equal(score, staged_scores[-1])
# AdaBoost regression
clf = AdaBoostRegressor(n_estimators=10)
clf.fit(boston.data, boston.target)
predictions = clf.predict(boston.data)
staged_predictions = [p for p in clf.staged_predict(boston.data)]
score = clf.score(boston.data, boston.target)
staged_scores = [s for s in clf.staged_score(boston.data, boston.target)]
assert_equal(len(staged_predictions), 10)
assert_array_almost_equal(predictions, staged_predictions[-1])
assert_equal(len(staged_scores), 10)
assert_array_almost_equal(score, staged_scores[-1])
def test_iris():
# Check consistency on dataset iris.
classes = np.unique(iris.target)
clf_samme = prob_samme = None
for alg in ['SAMME', 'SAMME.R']:
clf = AdaBoostClassifier(algorithm=alg)
clf.fit(iris.data, iris.target)
assert_array_equal(classes, clf.classes_)
proba = clf.predict_proba(iris.data)
if alg == "SAMME":
clf_samme = clf
prob_samme = proba
assert_equal(proba.shape[1], len(classes))
assert_equal(clf.decision_function(iris.data).shape[1], len(classes))
score = clf.score(iris.data, iris.target)
assert score > 0.9, "Failed with algorithm %s and score = %f" % \
(alg, score)
# Somewhat hacky regression test: prior to
# ae7adc880d624615a34bafdb1d75ef67051b8200,
# predict_proba returned SAMME.R values for SAMME.
clf_samme.algorithm = "SAMME.R"
assert_array_less(0,
np.abs(clf_samme.predict_proba(iris.data) - prob_samme))
def adaboost(X,training_target,Y,est):
from sklearn.ensemble import AdaBoostClassifier
clf = AdaBoostClassifier(n_estimators=est)
clf.fit(X,training_target)
proba = clf.predict_proba(Y)
def train(xTrain, yTrain, metric):
print 'adaboost'
global boost
boost = AdaBoostClassifier()
boost.fit(xTrain,yTrain)
global trainResults
trainResults = boost.predict_proba(xTrain)[:,1]
i.setSuccess(trainResults, metric)
def test_classification_toy():
# Check classification on a toy dataset.
for alg in ['SAMME', 'SAMME.R']:
clf = AdaBoostClassifier(algorithm=alg, random_state=0)
clf.fit(X, y_class)
assert_array_equal(clf.predict(T), y_t_class)
assert_array_equal(np.unique(np.asarray(y_t_class)), clf.classes_)
assert_equal(clf.predict_proba(T).shape, (len(T), 2))
assert_equal(clf.decision_function(T).shape, (len(T),))
def main():
X, Y = load('train.csv')
adaboost = AdaBoostClassifier(n_estimators=150, learning_rate=0.1)
adaboost.fit(X, Y)
X_test, ID = loadTest('test.csv')
target = adaboost.predict_proba(X_test)
df = pandas.DataFrame()
df['TARGET'] = target[:,1]
df.index = pandas.Series(ID, name='ID')
df.to_csv('sumbit.csv')
def abClassifier(X_train,y_train,X_test,y_test,to_plot=False):
params = {
'random_state': [None,0,1,2,3,4,5]}
for param in ParameterGrid(params):
print param
clf = AdaBoostClassifier(algorithm='SAMME.R',n_estimators=50,learning_rate=0.7,**param)
clf.fit(X_train,y_train)
# auc_compute(y_test,clf.predict_proba(X_test)[:,1])
predictions=clf.predict(X_test)
scores(y_test,predictions,clf.predict_proba(X_test)[:,1],'ab',to_plot=to_plot)
def classify_AdaBoost(train, test): from sklearn.ensemble import AdaBoostClassifier as ABC x, y = train clf = ABC() clf.fit(x, y) x, y = test proba = clf.predict_proba(x) return proba
class ABClassifier(Model):
'''
Adaptive Boosting Classifier
Boosting an initial result of 1-depth decision tress classifier
'''
def __init__(self):
Model.__init__(self)
self.model = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),algorithm="SAMME",n_estimators=200)
def predict(self, test):
return self.model.predict_proba(test)
def adaboost(X, y, train, valid):
from sklearn.ensemble import AdaBoostClassifier
clf2 = AdaBoostClassifier(n_estimators=100).fit(X[train], y[train])
yhat = clf2.predict(X[valid])
print(classification_report(y[valid], yhat))
accuracy_score(y[valid], yhat)
print("adaboost" + str(accuracy_score(y[valid], yhat)))
yhat_prob = clf2.predict_proba(X[valid])[:,1]
print("extra tree randomForest roc_accuracy" + str(roc_auc_score(y[valid], yhat_prob)))
np.savetxt("y_ada.csv", yhat_prob)
return yhat_prob
def test_multidimensional_X():
"""
Check that the AdaBoost estimators can work with n-dimensional
data matrix
"""
from sklearn.dummy import DummyClassifier, DummyRegressor
rng = np.random.RandomState(0)
X = rng.randn(50, 3, 3)
yc = rng.choice([0, 1], 50)
yr = rng.randn(50)
boost = AdaBoostClassifier(DummyClassifier(strategy='most_frequent'))
boost.fit(X, yc)
boost.predict(X)
boost.predict_proba(X)
boost = AdaBoostRegressor(DummyRegressor())
boost.fit(X, yr)
boost.predict(X)
def run_adaboost(estimators_and_learn_rt):
print estimators_and_learn_rt[0]
print estimators_and_learn_rt[1]
clf = AdaBoostClassifier(base_estimator=DecisionTreeClassifier(max_depth=1,
max_features=features-1,
splitter='best', min_samples_leaf=10),
n_estimators = int(estimators_and_learn_rt[0]),
learning_rate=estimators_and_learn_rt[1])
clf.fit(train_features, train_outcome)
validation['predictions_clf']=clf.predict_proba(validation_for_p)[:,1]
fpr, tpr, thresholds = roc_curve(validation.is_exciting, validation.predictions_clf)
auc_score = auc(fpr,tpr)
return auc_score
def AdaBoost(X, Y, XTest, YTest):
print '-----------------------------------------------------'
# param_grid = {'learning_rate': [0.1, 0.3, 0.6, 1, 3, 6, 10]}
# tree_grid = GridSearchCV(AdaBoostClassifier(), param_grid)
tree_grid = AdaBoostClassifier(n_estimators=100, learning_rate=2)
tree_grid.fit(X, Y)
# print("The best parameters are %s with a score of %0.2f"
# % (tree_grid.best_params_, tree_grid.best_score_))
print "Computing training statistics"
dtree_predict_time_training = time.time()
Ypred_dtree_training = tree_grid.predict(X)
dtree_predict_time_training = time.time() - dtree_predict_time_training
dtree_accuracy_training = metrics.accuracy_score(Y, Ypred_dtree_training)
dt_precision_training = metrics.precision_score(Y, Ypred_dtree_training,
average='binary')
dtree_recall_training = metrics.recall_score(Y, Ypred_dtree_training,
average='binary')
print "DT training prediction time: " + str(dtree_predict_time_training)
print "DT training accuracy Score: " + str(dtree_accuracy_training)
print "DT training precision Score: " + str(dt_precision_training)
print "DT training recall Score: " + str(dtree_recall_training)
print "Computing testing statistics"
dtree_predict_time_test = time.time()
Ypred_dtree_test = tree_grid.predict(XTest)
dtree_predict_time_test = time.time() - dtree_predict_time_test
dtree_accuracy_test = metrics.accuracy_score(YTest, Ypred_dtree_test)
dt_precision_test = metrics.precision_score(YTest, Ypred_dtree_test,
average='binary')
dtree_recall_test = metrics.recall_score(YTest, Ypred_dtree_test,
average='binary')
print "DT test prediction time: " + str(dtree_predict_time_test)
print "DT test accuracy Score: " + str(dtree_accuracy_test)
print "DT test precision Score: " + str(dt_precision_test)
print "DT test recall Score: " + str(dtree_recall_test)
print "Creating ROC curve"
y_true = YTest
y_score = tree_grid.predict_proba(XTest)
fprSVM, trpSVM, _ = metrics.roc_curve(y_true=y_true,
y_score=y_score[:, 0],
pos_label=0)
plt.plot(fprSVM, trpSVM, 'c-', label='ADA')
def iterate(self, n_estimators_conf=[10], learning_rate_conf=[0.25]): print '-'*80 print 'Running AdaBoost Iterations...' # performance by number of estimators and max depth results = [] for ne in n_estimators_conf: for lr in learning_rate_conf: print 'Iteration: n_estimators=%s, learning_rate=%s' % (str(ne), str(lr)) m = AB(n_estimators=ne, learning_rate=lr) m.fit(self.xtrain, self.ytrain) predtrain = m.predict(self.xtrain) predtest = m.predict(self.xtest) predprobatrain = m.predict_proba(self.xtrain) predprobatest = m.predict_proba(self.xtest) accuracytrain = metrics.accuracy_score(predtrain, self.ytrain) accuracytest = metrics.accuracy_score(predtest, self.ytest) kstrain = multiclass_log_loss(self.ytrain, predprobatrain) kstest = multiclass_log_loss(self.ytest, predprobatest) cr = self.convert_cr(metrics.classification_report(self.ytest, predtest)) results.append([ne, lr, accuracytrain, accuracytest, kstrain, kstest, cr]) self.results = pd.DataFrame(results) self.results.columns = ['ne', 'lr', 'accuracy_train', 'accuracy_test', 'ks_train', 'ks_test', 'cr']
def ada_boost_predict(new_train_data, new_train_labels, test_data, test_labels, base_est = "tree", n = 50):
# Create a classifier: AdaBoost classifier
if base_est == "tree":
base = DecisionTreeClassifier(max_depth=5)
classifier = AdaBoostClassifier(base_estimator = base, n_estimators = n)
# We learn the digits on the first half of the digits
classifier.fit(new_train_data, new_train_labels)
# Now predict the value of the digit on the second half:
expected = test_labels
predicted = classifier.predict_proba(test_data)
return predicted
def test_iris():
"""Check consistency on dataset iris."""
classes = np.unique(iris.target)
for alg in ['SAMME', 'SAMME.R']:
clf = AdaBoostClassifier(algorithm=alg)
clf.fit(iris.data, iris.target)
assert_array_equal(classes, clf.classes_)
assert_equal(clf.predict_proba(iris.data).shape[1], len(classes))
assert_equal(clf.decision_function(iris.data).shape[1], len(classes))
score = clf.score(iris.data, iris.target)
assert score > 0.9, "Failed with algorithm %s and score = %f" % \
(alg, score)
def main():
Algorithm = 'CamKt12LCTopoSplitFilteredMu67SmallR0YCut9'
print 'Loading training data ...'
data_train = pd.read_csv(Algorithm+'merged.csv')
r =np.random.rand(data_train.shape[0])
#Set label and weight vectors - and drop any unwanted tranining one
Y_train = data_train['label'].values[r<0.5]
# W_train = data_train['weight'].values[r<0.9]
Y_valid = data_train['label'].values[r>=0.5]
# W_valid = data_train['weight'].values[r>=0.9]
# data_train.drop('AKT10LCTRIM530_MassDropSplit', axis=1, inplace=True)
varcombinations = itertools.combinations(data_train.columns.values[1:-1],2)
fac = lambda n: 1 if n < 2 else n * fac(n - 1)
combos = lambda n, k: fac(n) / fac(k) / fac(n - k)
colors = plt.get_cmap('jet')(np.linspace(0, 1.0,combos(len(data_train.columns.values[1:-1]),2) ))
for varset,color in zip(varcombinations, colors):
print list(varset)
X_train = data_train[list(varset)].values[r<0.5]
X_valid = data_train[list(varset)].values[r>=0.5]
dt = DC(max_depth=3,min_samples_leaf=0.05*len(X_train))
abc = ABC(dt,algorithm='SAMME',
n_estimators=8,
learning_rate=0.5)
print 'Training classifier with all the data..'
abc.fit(X_train, Y_train)
print 'Done.. Applying to validation sample and drawing ROC'
prob_predict_valid = abc.predict_proba(X_valid)[:,1]
Y_score = abc.decision_function(X_valid)
fpr, tpr, _ = roc_curve(Y_valid, prob_predict_valid)
labelstring = ' And '.join(var.replace('_','') for var in varset)
print labelstring
plt.plot(tpr, (1-fpr), label=labelstring, color=color)
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
plt.ylabel('1- Background Efficiency')
plt.xlabel('Signal Efficiency')
plt.title(Algorithm+' ROC Curve')
plt.legend(loc="lower left",prop={'size':6})
plt.savefig(Algorithm+'rocmva.pdf')
def training(thistrainingfeatures, baseclassparameters, adaparameters):
treeclassifier = DecisionTreeClassifier(**baseclassparameters)
adaclassifier = AdaBoostClassifier(treeclassifier, **adaparameters)
#Split training and testdata
Xtrain, Xtest, ytrain, ytest = train_test_split(thistrainingfeatures, label)
#Cast pd.Series to arrays to apply mask later
ytrain = np.asarray(ytrain)
ytest = np.asarray(ytest)
#print "\nBegin calculation with {0} and {1}".format(str(baseclassparameters), str(adaparameters))
adaclassifier.fit(Xtrain, ytrain)
#Predict with the model
prob_predict_test = adaclassifier.predict_proba(Xtest)[:,1]
#Calculate maximal significance
True_Signal_test = prob_predict_test[ytest==1]
True_Bkg_test = prob_predict_test[ytest==0]
best_significance = 0
for x in np.linspace(0, 1, 1000):
S = float(len(True_Signal_test[True_Signal_test>x]))
B = float(len(True_Bkg_test[True_Bkg_test>x]))
significance = S/np.sqrt(S+B)
if significance > best_significance:
best_significance = significance
best_x = x
best_S = S
best_B = B
if best_significance > best_overall_significance:
print """\nCalculation with {0} and {1} done.
Variables: {2}
Best significance of {3:.2f} archived when cutting at {4:.3f}
Signal efficiency: {5:.2f}%
Background efficiency: {6:.2f}%
Purity: {7:.2f}%""".format( str(baseclassparameters), str(adaparameters),
str(list(thistrainingfeatures.columns)),
best_significance, best_x,
100.*best_S/len(True_Signal_test),
100.*best_B/len(True_Bkg_test),
100.*best_S/(best_S+best_B) )
#Print feature importances
for (feature, importance) in izip(thistrainingfeatures.columns, adaclassifier.feature_importances_):
print "{0:45s}: {1:>10.2f}%".format(feature, importance*100.)
class AdaBoostPredictor(PredictorBase):
'''
AdaBoost
'''
def __init__(self):
self.clf = AdaBoostClassifier()
def fit(self, X_train, y_train):
self.clf.fit(X_train, y_train)
def predict(self, X_test):
predictions = self.clf.predict_proba(X_test)
predictions_df = self.bundle_predictions(predictions)
return predictions_df
class ADAClassifier(Classifier, ProbClassifier):
def __init__(self, maxTreeDepth=1, estimators=50, learningRate=1.):
self.cl = AdaBoostClassifier(n_estimators=estimators, learning_rate=learningRate,
base_estimator=DecisionTreeClassifier(max_depth=maxTreeDepth))
def retrain(self, vectorFeature, vectorTarget):
# self.cl.fit([v.toarray()[0] for v in vectorFeature], vectorTarget)
self.cl.fit(vectorFeature, vectorTarget)
def classify(self, vectorizedTest):
# return self.cl.predict(vectorizedTest.toarray()[0])[0]
return self.cl.predict(vectorizedTest)[0]
def getProb(self, vectorizedTest):
# return self.cl.predict_proba(vectorizedTest.toarray()[0])[0][1]
return self.cl.predict_proba(vectorizedTest)[0][1]
def test(self):
X, y = self.dataMat,self.labelMat
X_test = self.testData
params = {'n_estimators': 1200, 'max_depth': 4, 'subsample': 0.5,'learning_rate': 0.01, 'min_samples_leaf': 1, 'random_state': 3}
#clf = GradientBoostingClassifier(**params)
clf=AdaBoostClassifier(DecisionTreeClassifier(max_depth=6),algorithm="SAMME.R",n_estimators=280)
clf.fit(X, y);
y_pred = clf.predict(X_test);
y_predprob = clf.predict_proba(X_test);
output.write('bidder_id'+','+'prediction'+'\n')
for i in range(0,len(self.totalid)):
if self.totalid[i] in self.testid:
idx = self.testid.index(self.totalid[i])
output.write(str(self.testid[idx])+','+str(y_predprob[idx][1])+'\n')
#print str(self.testid[idx])+','+str(y_predprob[idx][1])
else:
#print str(self.totalid[idx])+','+str(0.0)
output.write(str(self.totalid[i])+','+str(0.0)+'\n')
def test_sparse_classification():
# Check classification with sparse input.
class CustomSVC(SVC):
"""SVC variant that records the nature of the training set."""
def fit(self, X, y, sample_weight=None):
"""Modification on fit caries data type for later verification."""
super().fit(X, y, sample_weight=sample_weight)
self.data_type_ = type(X)
return self
X, y = datasets.make_multilabel_classification(n_classes=1, n_samples=15,
n_features=5,
random_state=42)
# Flatten y to a 1d array
y = np.ravel(y)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
for sparse_format in [csc_matrix, csr_matrix, lil_matrix, coo_matrix,
dok_matrix]:
X_train_sparse = sparse_format(X_train)
X_test_sparse = sparse_format(X_test)
# Trained on sparse format
sparse_classifier = AdaBoostClassifier(
base_estimator=CustomSVC(probability=True),
random_state=1,
algorithm="SAMME"
).fit(X_train_sparse, y_train)
# Trained on dense format
dense_classifier = AdaBoostClassifier(
base_estimator=CustomSVC(probability=True),
random_state=1,
algorithm="SAMME"
).fit(X_train, y_train)
# predict
sparse_results = sparse_classifier.predict(X_test_sparse)
dense_results = dense_classifier.predict(X_test)
assert_array_equal(sparse_results, dense_results)
# decision_function
sparse_results = sparse_classifier.decision_function(X_test_sparse)
dense_results = dense_classifier.decision_function(X_test)
assert_array_almost_equal(sparse_results, dense_results)
# predict_log_proba
sparse_results = sparse_classifier.predict_log_proba(X_test_sparse)
dense_results = dense_classifier.predict_log_proba(X_test)
assert_array_almost_equal(sparse_results, dense_results)
# predict_proba
sparse_results = sparse_classifier.predict_proba(X_test_sparse)
dense_results = dense_classifier.predict_proba(X_test)
assert_array_almost_equal(sparse_results, dense_results)
# score
sparse_results = sparse_classifier.score(X_test_sparse, y_test)
dense_results = dense_classifier.score(X_test, y_test)
assert_array_almost_equal(sparse_results, dense_results)
# staged_decision_function
sparse_results = sparse_classifier.staged_decision_function(
X_test_sparse)
dense_results = dense_classifier.staged_decision_function(X_test)
for sprase_res, dense_res in zip(sparse_results, dense_results):
assert_array_almost_equal(sprase_res, dense_res)
# staged_predict
sparse_results = sparse_classifier.staged_predict(X_test_sparse)
dense_results = dense_classifier.staged_predict(X_test)
for sprase_res, dense_res in zip(sparse_results, dense_results):
assert_array_equal(sprase_res, dense_res)
# staged_predict_proba
sparse_results = sparse_classifier.staged_predict_proba(X_test_sparse)
dense_results = dense_classifier.staged_predict_proba(X_test)
for sprase_res, dense_res in zip(sparse_results, dense_results):
assert_array_almost_equal(sprase_res, dense_res)
# staged_score
sparse_results = sparse_classifier.staged_score(X_test_sparse,
y_test)
dense_results = dense_classifier.staged_score(X_test, y_test)
for sprase_res, dense_res in zip(sparse_results, dense_results):
assert_array_equal(sprase_res, dense_res)
# Verify sparsity of data is maintained during training
types = [i.data_type_ for i in sparse_classifier.estimators_]
assert all([(t == csc_matrix or t == csr_matrix)
for t in types])
acc = cross_val_score(estimator = clf_ada, X = X_train, y = y_train, cv = cv, scoring='f1')
acc.mean(), acc.std()
# last step
clf_ada = AdaBoostClassifier(dt,
algorithm = 'SAMME',
n_estimators = 100,
learning_rate = 0.1,
random_state= 1337 )
clf_ada.fit(X_train, y_train)
y_pred = clf_ada.predict(X_test)
print(classification_report(y_test, y_pred))
y_pred = clf_ada.predict_proba(X_test)[:, 1]
roc_auc_score(y_test, y_pred)
#kf = KFold(n_splits = 5, random_state = 1337, shuffle = True)
rskf = RepeatedStratifiedKFold(n_splits=5, n_repeats=3, random_state=1337)
# Create arrays and dataframes to store results
auc_preds = []
test_preds = np.zeros(df_test.shape[0])
n_fold = 0
for idx_train, idx_valid in rskf.split(X_train, y_train):
train_x, train_y = X_train[idx_train], y_train[idx_train]
valid_x, valid_y = X_train[idx_valid], y_train[idx_valid]
df = pd.DataFrame(vals)
X = df.drop('Class', axis=1)
y = df.loc[:, 'Class']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.15, random_state=420)
#:# preprocessing
#:# model
model = AdaBoostClassifier(n_estimators=10, learning_rate=0.3)
model.fit(X_train, y_train)
#:# hash
#:# e37d46e1a5c0065376d1471f564f3ac7
md5 = hashlib.md5(str(model).encode('utf-8')).hexdigest()
print(f'md5: {md5}')
#:# audit
y_pred = model.predict(X_test)
y_pred_proba = model.predict_proba(X_test)[:, 1]
print(f'Accuracy: {model.score(X_test, y_test)}')
print(f'Area under ROC: {roc_auc_score(y_test, y_pred_proba)}')
#:# session info
sessionInfo = {
"python_version": python_version(),
"library_versions":[str(d) for d in pkg_resources.working_set]
}
with open('sessionInfo.txt', 'w') as f:
json.dump(sessionInfo, f, indent=4)
to_predict = [calendar[ii].lrank - calendar[ii].wrank, calendar[ii].welo-calendar[ii].lelo, calendar[ii].welosur-calendar[ii].lelosur, surf_into_num[calendar[ii].surface], \
round(wins_percent(*winner) - wins_percent(*loser), 3), round(wins_per_surface(*winner) - wins_per_surface(*loser), 3), \
round(av_first_serve(*winner) - av_first_serve(*loser), 3), round(av_first_serve_surface(*winner) - av_first_serve_surface(*loser), 3), \
round(av_second_serve(*winner) - av_second_serve(*loser), 3), round(av_second_serve_surface(*winner) - av_second_serve_surface(*loser), 3), \
round(av_first_return(*winner) - av_first_return(*loser), 3), round(av_first_return_surface(*winner) - av_first_return_surface(*loser), 3), \
round(av_second_return(*winner) - av_second_return(*loser), 3), round(av_second_return_surface(*winner) - av_second_return_surface(*loser), 3), \
round(av_aces(*winner) - av_aces(*loser), 3), round(av_aces_surface(*winner) - av_aces_surface(*loser), 3), \
round(av_dfs(*winner) - av_dfs(*loser), 5), round(av_dfs_surface(*winner) - av_dfs_surface(*loser), 5), \
round(av_bps(*winner) - av_bps(*loser), 3), round(av_bps_surface(*winner) - av_bps_surface(*loser), 3)]
clf.fit(X, Y)
X_test = to_predict
y_test = [1]
prediction = clf.predict([to_predict])
proba = clf.predict_proba([to_predict])
coeffs_test = [calendar[ii].cfw, calendar[ii].cfl]
q1 = 0.1*bank1
q2 = 0.1*bank2
q3 = 0.1*bank3
q4 = 0.1*bank4
'''print("Best params:", model.best_params_)
print("To predict:", to_predict)
print("Prediction:", prediction)
print("Probabilities:", proba)
print("testing:", "("+str(roi_1(prediction, q1, coeffs_test)), str(roi_2(prediction, q2, coeffs_test)), str(roi_3(prediction, q3, coeffs_test)), \
str(roi_4(prediction, proba, q4, coeffs_test))+")")'''
profit1 += roi_1(prediction, q1, coeffs_test)
data_test = pd.read_csv('adult.test',
header=None,
skiprows=1,
names=column_names)
for name in data_test.columns:
data_test[name] = pd.Categorical(data_test[name]).codes
x_test = data_test[data_test.columns[:-1]]
y_test = data_test[data_test.columns[-1]]
y_test_pred = model.predict(x_test)
print('测试集准确率：', accuracy_score(y_test, y_test_pred))
print('\t测试集查准率：', precision_score(y_test, y_test_pred))
print('\t测试集召回率：', recall_score(y_test, y_test_pred))
print('\t测试集F1：', f1_score(y_test, y_test_pred))
y_test_proba = model.predict_proba(x_test)
# print y_test_proba
y_test_proba = y_test_proba[:, 1]
fpr, tpr, thresholds = metrics.roc_curve(y_test, y_test_proba)
auc = metrics.auc(fpr, tpr)
print('AUC = ', auc)
# 或直接调用roc_auc_score
# print 'AUC = ', metrics.roc_auc_score(y_test, y_test_proba)
mpl.rcParams['font.sans-serif'] = 'SimHei'
mpl.rcParams['axes.unicode_minus'] = False
plt.figure(facecolor='w')
plt.plot(fpr, tpr, 'r-', lw=2, alpha=0.8, label='AUC=%.3f' % auc)
plt.plot((0, 1), (0, 1), c='b', lw=1.5, ls='--', alpha=0.7)
plt.xlim((-0.01, 1.02))
plt.ylim((-0.01, 1.02))
# Draw a horizontal barplot of importances_sorted
vImportancesSorted.plot(kind='barh', color='lightgreen')
plt.title('Features Importances')
plt.show()
# METHOD V: Boosting
# Ada Boosting
vAdaBoostClassifier = AdaBoostClassifier(base_estimator=vDecisionTree,
n_estimators=180,
random_state=SEED)
# Fit ada to the training set
vAdaBoostClassifier.fit(vXTrain, vYTrain)
# Compute the probabilities of obtaining the positive class
vYPredProba = vAdaBoostClassifier.predict_proba(vXTest)[:, 1]
vAdaROCAUC = roc_auc_score(vYTest, vYPredProba)
print('ROC AUC score: {:.2f}'.format(vAdaROCAUC))
# Gradient Boosting
vGradientBoostingClassifier = GradientBoostingClassifier(max_depth=4,
n_estimators=180,
random_state=SEED)
vGradientBoostingClassifier.fit(vXTrain, vYTrain)
vYPred = vGradientBoostingClassifier.predict(vXTest)
vRMSE = MSE(vYTest, vYPred)**(1 / 2)
print('Test set RMSE of Gradient Boosting Classifier: {:.2f}'.format(vRMSE))
# Stochastic Gradient Boosting
vStochasticGradientBoostingClassifier = GradientBoostingClassifier(
max_depth=4,
random_state=None)
classifier.fit(train_data, train_label)
tra_label = classifier.predict(train_data) # 训练集的预测标签
tes_label = classifier.predict(test_data) # 测试集的预测标签
print("训练集：", accuracy_score(train_label, tra_label))
print("测试集：", accuracy_score(test_label, tes_label))
matrix = confusion_matrix(test_label, tes_label, labels=[0, 1])
TP = matrix[1, 1]
TN = matrix[0, 0]
FP = matrix[0, 1]
FN = matrix[1, 0]
sn = TP / (TP + FN)
sp = TN / (TN + FP)
decision_score = classifier.predict_proba(test_data)
fprs, tprs, thresholds = roc_curve(test_label, decision_score[:, 1])
# plt.plot(fprs, tprs)
# plt.show()
roc_auc = auc(fprs, tprs)
plt.figure()
lw = 2
plt.plot(fprs,
tprs,
color='darkorange',
lw=lw,
label='ROC curve (area = %0.2f)' % roc_auc)
plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')
plt.xlim([0.0, 1.0])
plt.ylim([0.0, 1.05])
X_train, X_test, y_train, y_test = train_test_split(X, y,
test_size=0.3,
stratify=y,
random_state=SEED)
# Instantiate a classification-tree 'dt'
dt = DecisionTreeClassifier(max_depth=1, random_state=SEED)
# Instantiate an AdaBoostClassifier 'adb_clf'
adb_clf = AdaBoostClassifier(base_estimator=dt, n_estimators=100)
# Fit adb_clf to the training set
adb_clf.fit(X_train, y_train)
# Predict the test set probabilities of positive class
y_pred_proba = adb_clf.predict_proba(X_test)[:,1]
# Evaluate test roc_auc score
adb_clf_roc_auc_score = roc_auc_score(y_test, y_pred_proba)
# Print adb_clf_roc_auc score
print('ROC AUC score: {:.2f}'.format(adb_clf_roc_auc_score))
# Gradient Boosting (GB)
# Gradient Boosted Trees - sequential correction of predecessor's errors
# Does not tweak the weights of the training instances
# Fit each predictor is trained using its predecessor's residual errors as labels
# Gradient Boosted Trees - CART is used as a base learner
# Important parameter - shrinkage - prediction of each tree is shrinked after multiplication by a learning rate, eta (0 to 1)
# Similar to AdaBoost - trade-off between Eta and the number of estimators
# Decreasing learning rate, needs to be compensated by increasing the number of estimators
print(classification_report(y_test, rf.predict(X_test)))
#print('Accuracy of Random Forest Classifier on test set: {:.2f}'.format(rf.score(X_test, y_test)*100))
# Classification report for the optimised RF Regression
rf.fit(X_train, y_train)
rfp = rf.predict(X_test)
#adaboost model
ada = AdaBoostClassifier(n_estimators=100, random_state=0)
ada.fit(X_train, y_train)
print("AdaBoost accuracy is %2.2f" %
accuracy_score(y_test, ada.predict(X_test)))
ada_roc_auc = roc_auc_score(y_test, ada.predict(X_test))
print("AdaBoost AUC = %2.2f" % ada_roc_auc)
######probability leaves employee
probsada = ada.predict_proba(
X_test)[:, 1] # predict probabilities associated with the employee leaving
adaProb_roc_auc = roc_auc_score(
y_test, probsada) # calculate AUC score using test dataset
print('AUC score: %.3f' % adaProb_roc_auc)
print(classification_report(y_test, ada.predict(X_test)))
#decision tree model
dtree = tree.DecisionTreeClassifier(max_depth=3,
class_weight="balanced",
min_weight_fraction_leaf=0.01)
dtree.fit(X_train, y_train)
print("Decision Tree accuracy is %2.2f" %
accuracy_score(y_test, dtree.predict(X_test)))
dt_roc_auc = roc_auc_score(y_test, dtree.predict(X_test))
print("Decision Tree AUC = %2.2f" % dt_roc_auc)
def train_bdt_multiclass():
print("Loading data...")
if SMALL_DATA:
signal, bkg2nu, bkg214Bi, bkg208Tl, bkgRn = import_data_small()
else:
signal, bkg2nu, bkg214Bi, bkg208Tl, bkgRn = import_data()
print("Creating arrays...")
# X = Features (i.e. the data)
X = np.concatenate((signal, bkg2nu, bkg214Bi, bkg208Tl, bkgRn))
# y = Labels (i.e. what it is, signal / background)
y = np.concatenate((np.ones(signal.shape[0]), np.full(bkg2nu.shape[0], 2),
np.full(bkg214Bi.shape[0],
3), np.full(bkg208Tl.shape[0],
4), np.full(bkgRn.shape[0], 5)))
print("Splitting Data...")
# Split the data
X_dev, X_eval, y_dev, y_eval = train_test_split(X,
y,
test_size=0.33,
random_state=48)
X_train, X_test, y_train, y_test = train_test_split(X,
y,
test_size=0.33,
random_state=42)
print("Creating classifier for DT")
# Create classifiers
dt = DecisionTreeClassifier(max_depth=12,
min_samples_split=0.5,
min_samples_leaf=400)
print("Creating classifier for BDT")
bdt = AdaBoostClassifier(dt,
algorithm='SAMME',
n_estimators=1200,
learning_rate=0.5)
print("Fitting BDT...")
# Train the classifier - not using weights here as it is a multiclassifier
fitted_tree = bdt.fit(X_train, y_train)
print("Predicting on training data...")
# Use the fitted tree to predict on training data and new test data
y_predicted_train = bdt.predict(X_train)
print("Predicting on test data...")
y_predicted_test = bdt.predict(X_test)
print(
classification_report(
y_train,
y_predicted_train,
target_names=["signal", "2nu", "214Bi", "208Tl", "Radon"]))
print("Area under ROC curve for training data: {0:.4f}".format(
roc_auc_score(y_train,
bdt.predict_proba(X_train),
average="weighted",
multi_class="ovr")))
print(
classification_report(
y_test,
y_predicted_test,
target_names=["signal", "2nu", "214Bi", "208Tl", "Radon"]))
print("Area under ROC curve for test data: {0:.4f}".format(
roc_auc_score(y_test,
bdt.predict_proba(X_test),
average="weighted",
multi_class="ovr")))
plot_roc_curve(bdt, X_test, y_test)
compare_train_test_multi(bdt, X_train, y_train, X_test, y_test)
print("Saving classifier...")
save_path = BASE_PATH + 'ml_calculated_data/multiClass/'
dump(bdt, save_path + 'bdt_classifier.joblib')
dump(fitted_tree, save_path + 'bdt_fitted_tree.joblib')
dump(X_train, save_path + 'bdt_X_train.joblib')
dump(X_test, save_path + 'bdt_X_test.joblib')
dump(X_dev, save_path + 'bdt_X_dev.joblib')
dump(X_eval, save_path + 'bdt_X_eval.joblib')
dump(y_test, save_path + 'bdt_y_test.joblib')
dump(y_train, save_path + 'bdt_y_train.joblib')
dump(y_dev, save_path + 'bdt_y_dev.joblib')
dump(y_eval, save_path + 'bdt_y_eval.joblib')
print("Finished Training.")
def AdaMECvsAdaBoost(dataset, C_FP, C_FN, base_estimator, algorithm,
n_estimators, calibration_method, test_set_prcnt,
cal_set_prcnt):
## Load data
mat_contents = sio.loadmat(os.getcwd() + '\\Datasets\\' + dataset + '.mat')
data = mat_contents['data']
target = np.asarray([float(i) for i in mat_contents['labels'].ravel()])
target[np.where(target != 1)] = 0 # One-vs-all if multiclass
## Split the data into training and test sets
X_train, X_test, y_train, y_test = train_test_split(
data, target, test_size=test_set_prcnt)
Pos = sum(
y_train[np.where(y_train == 1)]
) #Number of positive training examples --estimate of prior of positive class
Neg = len(
y_train
) - Pos #Number of negative training examples --estimate of prior of negative class
C_FP_effective = C_FP * Neg / (
C_FN * Pos + C_FP * Neg
) #Positive skew (overall importance of a single positive example)
#C_FN_effective = 1 - C_FP_effective #Negative skew (overall importance of a single negative example)
#Define weak learner
base_estimator = eval(base_estimator)
## Train ensembles
#I.Train an AdaBoost ensemble (algorithm="SAMME" for discrete AdaBoost, algorithm="SAMME.R" for real AdaBoost)
AdaBoost = AdaBoostClassifier(base_estimator,
algorithm=algorithm,
n_estimators=n_estimators)
AdaBoost = AdaBoost.fit(X_train, y_train)
#II.Train a Calibrated AdaBoost ensemble
AdaBoostCal = CalibratedAdaMEC.trainCalibratedAdaMEC(
base_estimator, algorithm, n_estimators, calibration_method,
cal_set_prcnt, X_train, y_train)
## Generate predictions
#I.AdaBoost predictions and scores
scores_AdaBoost = AdaBoost.predict_proba(X_test)[:,
1] #Positive Class scores
y_pred_AdaBoost = np.zeros(X_test.shape[0])
y_pred_AdaBoost[np.where(
scores_AdaBoost > 0.5
)] = 1 #Classifications, the standard AdaBoost decision rule corresponds to a threshold of 0.5 (skew-insensitive)
#II.Calibrated AdaMEC predictions and scores
y_pred_CalibratedAdaMEC, scores_CalibratedAdaMEC = CalibratedAdaMEC.predictCalibratedAdaMEC(
AdaBoostCal, C_FP_effective, X_test)
##Print results: comment/uncomment to your liking!
# #Confusion matrices
#print('AdaBoost Confusion Matrix:')
#conf_mat_AdaBoost = metrics.confusion_matrix(y_test, y_pred_AdaBoost)
#print(conf_mat_AdaBoost)
#print('Calibrated AdaMEC Confusion Matrix:')
#conf_mat_CalibratedAdaMEC = metrics.confusion_matrix(y_test, y_pred_CalibratedAdaMEC)
#print(conf_mat_CalibratedAdaMEC)
# #Accuracy (lower means better *skew-insensitive* classification).
# Note: Not a good measure for *skew-sensitive* learning.
#print('Accuracy:')
#print('\t\t\tAdaBoost: {0}'.format(metrics.accuracy_score(y_test, y_pred_AdaBoost)))
#print('\t\t\tCalibrated AdaMEC: {0}'.format(metrics.accuracy_score(y_test, y_pred_CalibratedAdaMEC)))
#Brier Score (lower means better probability estimates)
print('Brier Score:')
print('\t\t\tAdaBoost: {0}'.format(
metrics.brier_score_loss(y_test, scores_AdaBoost)))
print('\t\t\tCalibrated AdaMEC: {0}'.format(
metrics.brier_score_loss(y_test, scores_CalibratedAdaMEC)))
#Negative Log-likelihood (lower means better probability estimates)
print('Negative Log-likelihood:')
print('\t\t\tAdaBoost: {0}'.format(
metrics.log_loss(y_test, scores_AdaBoost)))
print('\t\t\tCalibrated AdaMEC: {0}'.format(
metrics.log_loss(y_test, scores_CalibratedAdaMEC)))
#Misclassification Cost (lower means better skew-sensitive classification)
print('Misclassification Cost:')
conf_mat_AdaBoost = metrics.confusion_matrix(
y_test, y_pred_AdaBoost) #Confusion matrix
cost_AdaBoost = conf_mat_AdaBoost[
0, 1] * C_FP_effective + conf_mat_AdaBoost[1, 0] * (
1 - C_FP_effective) #Skew-Sensitive Cost
print('\t\t\tAdaBoost: {0}'.format(cost_AdaBoost))
conf_mat_CalibratedAdaMEC = metrics.confusion_matrix(
y_test, y_pred_CalibratedAdaMEC) #Confusion matrix
cost_AdaMEC = conf_mat_CalibratedAdaMEC[
0, 1] * C_FP_effective + conf_mat_CalibratedAdaMEC[1, 0] * (
1 - C_FP_effective) #Skew-Sensitive Cost
print('\t\t\tCalibrated AdaMEC: {0}'.format(cost_AdaMEC))
if cost_AdaBoost > cost_AdaMEC:
print('Calibrated AdaMEC outperformed AdaBoost!')
else:
print('AdaBoost produced a lower cost solution this time. Try again.')
print('Calibrated AdaMEC should lead to lower cost in expectation.')
algorithm="SAMME",
n_estimators=number_of_estimators_all_attr,
learning_rate=rate_of_learning_all_attr)
print 'fitting bdt...'
ti = timer()
bdt.fit(X_train, y_train, sample_weight=w_train)
clf.fit(X_train, y_train, sample_weight=w_train)
tf = timer()
print 'bdt fit completed>>>>>>>>>>>>>>>>>>>>>>>'
print 'time taken for bdt fit1: ' + str(tf - ti) + 'sec'
#joblib.dump(bdt,'bdt.pkl')
#bdt = joblib.load('bdt.pkl')
#~~~~~~~~calculate the decision scores
#twoclass_output = bdt.decision_function(X_train)
all_probs = bdt.predict_proba(X_test)
dftt = df_test_orig.copy()
class_names = {0: "background", 1: "signal"}
classes = sorted(class_names.keys())
for cls in classes:
dftt[class_names[cls]] = all_probs[:, cls]
sig = dftt[isSigL] == 1
bkg = dftt[isSigL] == 0
probs = dftt["signal"][sig].values
probb = dftt["signal"][bkg].values
es, eb = [], []
for c in np.arange(-1, 1, roc_resolution):
es.append((float((probs > c).sum()) / probs.size))
eb.append((float((probb > c).sum()) / probb.size))
def create_model(dataset):
print("dataset : ", dataset)
df = pd.read_csv('/home/farshid/Desktop/' + dataset, header=None)
print('reading', dataset)
df['label'] = df[df.shape[1] - 1]
#
df.drop([df.shape[1] - 2], axis=1, inplace=True)
labelencoder = LabelEncoder()
df['label'] = labelencoder.fit_transform(df['label'])
#
X = np.array(df.drop(['label'], axis=1))
y = np.array(df['label'])
number_of_clusters = 23
sampler = RandomUnderSampler()
normalization_object = Normalizer()
X = normalization_object.fit_transform(X)
skf = StratifiedKFold(n_splits=5, shuffle=True)
n_classes = 2
for train_index, test_index in skf.split(X, y):
X_train = X[train_index]
X_test = X[test_index]
y_train = y[train_index]
y_test = y[test_index]
break
print('training', dataset)
top_roc = 0
depth_for_rus = 0
split_for_rus = 0
for depth in range(3, 20, 20):
for split in range(3, 9, 20):
classifier = AdaBoostClassifier(DecisionTreeClassifier(
max_depth=depth, min_samples_split=split),
n_estimators=100,
learning_rate=1,
algorithm='SAMME')
X_train, y_train = sampler.fit_sample(X_train, y_train)
classifier.fit(X_train, y_train)
predictions = classifier.predict_proba(X_test)
score = roc_auc_score(y_test, predictions[:, 1])
if top_roc < score:
top_roc = score
tpr = dict()
fpr = dict()
roc = dict()
for i in range(n_classes):
fpr[i], tpr[i], _ = roc_curve(y_test, predictions[:, i])
roc[i] = roc_auc_score(y_test, predictions[:, i])
major_class = max(sampler.fit(X_train, y_train).stats_c_,
key=sampler.fit(X_train, y_train).stats_c_.get)
major_class_X_train = []
major_class_y_train = []
minor_class_X_train = []
minor_class_y_train = []
for index in range(len(X_train)):
if y_train[index] == major_class:
major_class_X_train.append(X_train[index])
major_class_y_train.append(y_train[index])
else:
minor_class_X_train.append(X_train[index])
minor_class_y_train.append(y_train[index])
# optimize for number of clusters here
kmeans = KMeans(max_iter=200, n_jobs=4, n_clusters=number_of_clusters)
kmeans.fit(major_class_X_train)
# get the centroids of each of the clusters
cluster_centroids = kmeans.cluster_centers_
# get the points under each cluster
points_under_each_cluster = {
i: np.where(kmeans.labels_ == i)[0]
for i in range(kmeans.n_clusters)
}
for i in range(number_of_clusters):
size = len(points_under_each_cluster[i])
random_indexes = np.random.randint(low=0,
high=size,
size=int(size / 2))
temp = points_under_each_cluster[i]
feature_indexes = temp[random_indexes]
X_train_major = np.concatenate(
(X_train_major, X_train[feature_indexes]), axis=0)
y_train_major = np.concatenate(
(y_train_major, y_train[feature_indexes]), axis=0)
final_train_x = np.concatenate((X_train_major, minor_class_X_train),
axis=0)
final_train_y = np.concatenate((y_train_major, minor_class_y_train),
axis=0)
classifier = AdaBoostClassifier(DecisionTreeClassifier(max_depth=150))
# classifier = sklearn.svm.SVC(C=50 , gamma= .0008 , kernel='rbf', probability=True)
# classifier = sklearn.svm.SVC(C=100, gamma=.006, kernel='rbf', probability=True)
classifier.fit(final_train_x, final_train_y)
predicted = classifier.predict_proba(X_test)
tpr_c = dict()
fpr_c = dict()
roc_c = dict()
for i in range(n_classes):
fpr_c[i], tpr_c[i], _ = roc_curve(y_test, predictions[:, i])
roc_c[i] = auc(y_test, predictions[:, i])
print('ploting', dataset)
# plt.clf()
plt.plot(fpr[1],
tpr[1],
lw=2,
color='red',
label='Roc curve: Clustered sampling')
plt.plot(fpr_c[1],
tpr_c[1],
lw=2,
color='navy',
label='Roc curve: random under sampling')
plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')
plt.xlabel('Recall')
plt.ylabel('Precision')
plt.ylim([0.0, 1.05])
plt.xlim([0.0, 1.0])
plt.title('Area under ROC curve')
plt.legend(loc="lower right")
plt.show()
for index in sortedIndicies.tolist():
if classLabels[index] == 1.0:
delX = 0
delY = yStep
else:
delX = xStep
delY = 0
ySum += cursor[1]
# draw line from cursor to (cursor[0]-delX,cursor[1]-delY)
ax.plot([cursor[0], cursor[0] - delX], [cursor[1], cursor[1] - delY],
c='b')
cursor = (cursor[0] - delX, cursor[1] - delY)
ax.plot([0, 1], [0, 1], 'b--')
plt.xlabel('False positive rate')
plt.ylabel('True positive rate')
plt.title('ROC cursorve for AdaBoost horse colic detection system')
ax.axis([0, 1, 0, 1])
plt.show()
# 每个小矩形相加，矩形的宽度为xStep，因此对矩形的高度进行相加得到ySum
print("the Area Under the cursorve is: ", ySum * xStep)
if __name__ == "__main__":
X, y = make_hastie_10_2(n_samples=4000, random_state=1)
X_test, y_test = X[2000:], y[2000:]
X_train, y_train = X[:2000], y[:2000]
clf = AdaBoostClassifier(n_estimators=100)
clf.fit(X_train, y_train)
preds = clf.predict_proba(X_test)
plotROC(preds[:, 1], y_test)
#NOTE: change classifier here
clf = AdaBoostClassifier(n_estimators=500, algorithm='SAMME')
#training
st = time.time()
print "training started"
clf.fit(x_train, y_train)
print "training ended"
et = time.time()
tt = et - st
print "Training Time = " + str(tt) + "\n"
#predictions
pred = clf.predict(x_test)
#NOTE: change to decision_function or predict_proba depending on the classifier
y_score = clf.predict_proba(x_test)
#y_score = clf.decision_function(x_test)
#################################################################################
pp = PdfPages('results/EXP_Result.pdf')
#PrecisionRecall-plot
precision = dict()
recall = dict()
PR_area = dict()
PR_thresholds = dict()
average_precision = dict()
for i in range(n_classes):
precision[i], recall[i], PR_thresholds[i] = precision_recall_curve(
y_test[:, i], y_score[:, i])
PR_area[i] = auc(recall[i], precision[i])
average_precision[i] = average_precision_score(y_test[:, i], y_score[:, i])
plt.ylabel('Test Accuracy%')
plt.xlabel('n_estimators')
plt.show()
# ROC curve for baseline classification tree
clf_probs=clf.predict_proba(wine_test.loc[:,['price','regn_enc','var_enc', \
'wnry_enc']])
fpr1,tpr1,thr1=roc_curve(np.where(wine_test['point_bins']=='90+',1.,0.), \
clf_probs[:,0])
# ROC curve for bagging ensemble using full classification trees
bag_probs=baglfy.predict_proba(wine_test.loc[:,['price','regn_enc', \
'var_enc','wnry_enc']])
fpr2,tpr2,thr2=roc_curve(np.where(wine_test['point_bins']=='90+',1.,0.), \
bag_probs[:,0])
# ROC curve for boosting ensemble using full classification trees
bst_probs=bstlfy.predict_proba(wine_test.loc[:,['price','regn_enc', \
'var_enc','wnry_enc']])
fpr3,tpr3,thr3=roc_curve(np.where(wine_test['point_bins']=='90+',1.,0.), \
bst_probs[:,0])
# Plot ROC Curves
plt.plot(fpr1,tpr1,color='#4d4d33',label='Baseline CART')
plt.plot(fpr2,tpr2,color='#0080ff',label='Bagging Ensemble')
plt.plot(fpr3,tpr3,color='#ff3300',label='Boosting Ensemble')
plt.plot([0.,1.],[0.,1.],color='k',linestyle='--')
plt.title('ROC Curves for 90+ Point Wine Classification')
plt.xlabel('false positive rate')
plt.ylabel('true positive rate')
plt.legend(fontsize=8)
plt.show()
y_pred_dtc = dtc.predict(X_test_prepared0)
print("dtc percentage: ", 100 * np.sum(y_pred_dtc == Val_y) / len(Val_y))
y_score_sgd = f1_score(Val_y, y_pred_sgd)
print("sgd f1 score: ", y_score_sgd)
y_score_gbc = f1_score(Val_y, y_pred_gbc)
print("gbc f1 score: ", y_score_gbc)
y_score_adb = f1_score(Val_y, y_pred_adb)
print("adb f1 score: ", y_score_adb)
y_score_dtc = f1_score(Val_y, y_pred_dtc)
print("dtc f1 score: ", y_score_dtc)
auc_sgd = roc_auc_score(Val_y, y_pred_sgd)
auc_gbc = roc_auc_score(Val_y, y_pred_gbc)
auc_adb = roc_auc_score(Val_y, y_pred_adb)
auc_dtc = roc_auc_score(Val_y, y_pred_dtc)
print("sgd auc: ", auc_sgd)
print("gbc auc: ", auc_gbc)
print("adb auc: ", auc_adb)
print("dtc auc: ", auc_dtc)
# y_sgd_predict = sgd.predict_proba(X_test_prepared)
y_gbc_predict = gbc.predict_proba(X_test_prepared)
y_adb_predict = adb.predict_proba(X_test_prepared)
y_dtc_predict = dtc.predict_proba(X_test_prepared)
np.save("y_gbc_predict", y_gbc_predict)
np.save("y_adb_predict", y_adb_predict)
np.save("y_dtc_predict", y_dtc_predict)
#KNN
from sklearn.neighbors import KNeighborsClassifier
rf6 = KNeighborsClassifier()
rf6.fit(X_train, y_train)
y_val_pred6 = rf6.predict_proba(X_val)
y_val_pred_acc6 = rf6.predict(X_val)
print(log_loss(y_val, y_val_pred6))
print(accuracy_score(y_val, y_val_pred_acc6))
#AdaBoost
from sklearn.ensemble import AdaBoostClassifier
rf7 = AdaBoostClassifier(n_estimators=250)
rf7.fit(X_train, y_train)
y_val_pred6 = rf7.predict_proba(X_val)
y_val_pred_acc7 = rf7.predict(X_val)
print(log_loss(y_val, y_val_pred7))
print(accuracy_score(y_val, y_val_pred_acc7))
#Compare ROC of each Algorithm
import matplotlib.pyplot as plt
from sklearn import metrics
#RandomForest
fpr1, tpr1, threshold1 = metrics.roc_curve(y_val_pred_acc1, y_val_pred1)
roc_auc1 = metrics.auc(fpr1, tpr1)
plt.title('ROC of RandomForest')
plt.plot(fpr1, tpr1, 'b', label='AUC = %0.2f' % roc_auc1)
plt.legend(loc='lower right')
plt.plot([0, 1], [0, 1], 'r--')
plt.xlim([0, 1])
def main(Testbatch, DocumentName):
start_time = time.time()
Indexlist = pd.read_csv('Index.csv', sep=";", header=None)
Indexlist.columns = ['Index', 'document']
Dataset = 'ENRON' #'TREC'
catR = 'Fraud' #'spam'
catNR = 'Legit' #'ham'
#%%
NrMails = 1640 #75000/5
SavePer = 40 #100
Traininglist = pd.read_csv('Traininglist'+str(Dataset),sep = '\t', index_col = 0)
wordselection =pd.read_csv('wordselection-'+str(Dataset)+str(DocumentName), sep = '\t', index_col = 0, names=['Words'])
#%%
# =============================================================================
# start of training LOGISTIC REGRESSION
# =============================================================================
Training = pd.DataFrame(0, columns = wordselection, index = [], dtype = 'uint32')
y = pd.DataFrame(0, columns = [], index = [], dtype = 'uint32')
Batches = list(Traininglist.columns.values)
Batches.remove(Testbatch)
for batch in Batches:
print(batch)
for Files in range(int(SavePer),int(NrMails+SavePer),int(SavePer)):
TrainingFile = pd.DataFrame()
TrainingFile = pd.read_csv('Frequencies'+str(Dataset)+batch+'-'+str(Files),sep = '\t', index_col = 0)
y = pd.concat([y, TrainingFile['Index_given']], sort=False, ignore_index=True)
Training = pd.concat([Training, TrainingFile], sort=False, ignore_index=True)
Training = Training[wordselection]
Training = Training.fillna(0).to_sparse(fill_value=0)
print(round(Files/float(NrMails)*100,4), '%')
del TrainingFile
Training.to_csv('Training'+str(Dataset)+str(DocumentName),sep='\t')
train = AdaBoostClassifier(n_estimators = 100, random_state=0).fit(Training, y)
#%%
# =============================================================================
# Applying LOGISTIC REGRESSION to test data
# =============================================================================
start_time2 = time.time()
TrainingWords = list(Training.columns.values)
Test = pd.DataFrame(0, columns = TrainingWords, index = [], dtype = 'uint32')
Traininglist = pd.read_csv('Traininglist'+str(Dataset),sep = '\t', index_col = 0)
Batches = list(Traininglist.columns.values)
Batches.remove(Testbatch)
ProbSpam = list()
ProbHam = list()
Given_y = list()
Predicted_y = list()
for Files in range(int(SavePer),int(NrMails+SavePer),int(SavePer)):
Test = pd.DataFrame(0, columns = TrainingWords, index = [], dtype = 'uint32')
TestFile = pd.read_csv('Frequencies'+str(Dataset)+batch+'-'+str(Files),sep = '\t', index_col=0)
y = TestFile['Index_given']
del TestFile['Index_given']
Test = Test.merge(TestFile, how='outer')
for word in list(set(Test.columns.values)-set(TrainingWords)):
del Test[word]
TestNew = Test.fillna(0).to_sparse(fill_value=0)
pred = train.predict(TestNew)
proba = train.predict_proba(TestNew)
for i in range(0,len(Test)):
ProbSpam.append(proba[i][0])
ProbHam.append(proba[i][1])
Given_y.append(y[i])
Predicted_y.append(pred[i])
print(round(Files/float(NrMails)*100,4), '%')
Result = pd.DataFrame(0, columns = ["Given_Label", "Predicted_Label", "ProbSpam", "ProbHam", "expSpam", "expHam"], index = [], dtype = 'uint32')
Result["Given_Label"] = Given_y
Result["Predicted_Label"] = Predicted_y
Result["ProbSpam"] = ProbSpam
Result["ProbHam"] = ProbHam
Result["expSpam"] = 0
Result["expHam"] = 0
Result.to_csv('Result'+str(Dataset)+'-'+str(DocumentName),sep='\t')
Timings = pd.DataFrame(columns = ['Description','time'])
Timings = Timings.append(pd.Series({'Description':'Script', 'time': time.time()-start_time}), ignore_index=True)
Timings = Timings.append(pd.Series({'Description':'Training', 'time': start_time2-start_time}), ignore_index=True)
Timings = Timings.append(pd.Series({'Description':'Classification', 'time': time.time()- start_time2}), ignore_index=True)
Timings.to_csv('Timings'+str(Dataset)'-'+str(DocumentName),sep='\t')
pred = rf_model.predict_proba(X_train[training_vars])
print('RF train roc-auc: {}'.format(roc_auc_score(y_train, pred[:,1])))
pred = rf_model.predict_proba(X_test[training_vars])
print('RF test roc-auc: {}'.format(roc_auc_score(y_test, pred[:,1])))
# #### Adaboost
# In[283]:
ada_model = AdaBoostClassifier()
ada_model.fit(X_train[training_vars], y_train)
pred = ada_model.predict_proba(X_train[training_vars])
print('Adaboost train roc-auc: {}'.format(roc_auc_score(y_train, pred[:,1])))
pred = ada_model.predict_proba(X_test[training_vars])
print('Adaboost test roc-auc: {}'.format(roc_auc_score(y_test, pred[:,1])))
# #### Logistic Regression
# In[4]:
logit_model = LogisticRegression()
logit_model.fit(scaler.transform(X_train[training_vars]), y_train)
pred = logit_model.predict_proba(scaler.transform(X_train[training_vars]))
print('Logit train roc-auc: {}'.format(roc_auc_score(y_train, pred[:,1])))
def adaBoostClassifier(xTrain, yTrain, xTest):
adaClassifier = AdaBoostClassifier()
adaClassifier.fit(xTrain, yTrain)
yPredict = adaClassifier.predict(xTest)
probability = adaClassifier.predict_proba(xTest)
return yPredict, probability
from sklearn.ensemble import AdaBoostClassifier
from sklearn.datasets import make_moons, make_circles, make_classification
#引入训练数据
#X, y = make_circles(noise=0.2, factor=0.5, random_state=1)
X, y = make_moons(noise=0.1, random_state=1)
#定义AdaBoost分类器
adb = AdaBoostClassifier()
#训练过程
adb.fit(X, y)
#绘图库引入
import matplotlib.pyplot as plt
import matplotlib as mpl
import numpy as np
#调整图片风格adbadb
mpl.style.use('fivethirtyeight')
#定义xy网格，用于绘制等值线图
x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5
xx, yy = np.meshgrid(np.arange(x_min, x_max, 0.1),
np.arange(y_min, y_max, 0.1))
#预测可能性
Z = adb.predict_proba(np.c_[xx.ravel(), yy.ravel()])[:, 1]
Z = Z.reshape(xx.shape)
plt.contourf(xx, yy, Z, alpha=.8)
#绘制散点图
plt.scatter(X[:, 0], X[:, 1], c=y, edgecolors='k')
plt.title("AdaBoost")
plt.axis("equal")
plt.show()
##tfidf = feature_extraction.text.TfidfTransformer()
##train_data = tfidf.fit_transform(train_data).toarray()
##test_data = tfidf.transform(test_data).toarray()
print 'Training...'
forest = GradientBoostingClassifier(n_estimators=200, verbose=1,
learning_rate = 0.2, max_depth=3)
forest2 = RandomForestClassifier(n_estimators = 400, verbose = 1,
max_features = 13)
learner = AdaBoostClassifier(base_estimator = forest2, n_estimators = 50)
forest = forest.fit(train_data, y)
learner = learner.fit(train_data, y)
print 'Predicting...'
output1 = forest.predict_proba(test_data)
output2 = learner.predict_proba(test_data)
output = []
for t, row in enumerate(output1):
tmp = np.vstack([output1[t], output2[t]])
tmp = np.average(tmp, axis = 0)
output.append(tmp)
output = np.array(output)
predictions_file = open("submission.csv", "wb")
open_file_object = csv.writer(predictions_file)
open_file_object.writerow(['Id', 'Class_1', 'Class_2', 'Class_3',
'Class_4', 'Class_5', 'Class_6',
'Class_7', 'Class_8', 'Class_9'])
for t, row in enumerate(Ids):
# Gradient Boosting classifier
gbc = GradientBoostingClassifier(n_estimators = 200, learning_rate = 0.05, random_state=12)
gbc.fit(x_train,y_train)
# **Best solo performer from above is AdaBoost with AUROC of ~0.8612. A close second was the MLP, with AUROC of ~0.8574
# on the validation set.
predictions_LR_train = logreg.predict_proba(x_2)[:,1]
predictions_DT_train = dt.predict_proba(x_2)[:,1]
predictions_NN_train = nn.predict_proba(x_2)[:,1]
predictions_GBC_train = gbc.predict_proba(x_2)[:,1]
predictions_KNN_train = knn.predict_proba(x_2)[:,1]
predictions_RF_train = rf.predict_proba(x_2)[:,1]
predictions_AB_train = ab.predict_proba(x_2)[:,1]
predictions_GNB_train = gnb.predict_proba(x_2)[:,1]
# Reshape to get the arrays to work
predictions_LR_train = predictions_LR_train.reshape(-1, 1)
predictions_DT_train = predictions_DT_train.reshape(-1, 1)
predictions_NN_train = predictions_NN_train.reshape(-1, 1)
predictions_GBC_train = predictions_GBC_train.reshape(-1, 1)
predictions_KNN_train = predictions_KNN_train.reshape(-1, 1)
predictions_RF_train = predictions_RF_train.reshape(-1, 1)
predictions_AB_train = predictions_AB_train.reshape(-1, 1)
predictions_GNB_train = predictions_GNB_train.reshape(-1, 1)
# What to train the meta model on
next_x_train = np.concatenate((predictions_LR_train,predictions_DT_train, predictions_NN_train, predictions_KNN_train,
predictions_RF_train, predictions_AB_train, predictions_GNB_train,
with open('resultAda.csv', 'w') as csvFile:
writer = csv.writer(csvFile, delimiter=' ')
writer.writerows(predict)
dfAda = predict
csvFile.close()
uniciAda, counteggioAda = np.unique(dfAda, return_counts=True)
print(uniciAda, counteggioAda)
print("\nRilevanza attributi Ada")
for nameAda, scoreAda in zip(COLUMNS, classificatore.feature_importances_):
print(nameAda, scoreAda)
# testo il classificatore con delle classi che già conosce e di cui conosce anche l array di feature per testarne la precisione
predict_proba = classificatore.predict_proba(dataframe_training)
predict = np.array(predict)
classi_target = list(np.array(classi_target))
cnf_matrix = confusion_matrix(classi_target, predict)
print(' - Confusion Matrix -')
print(cnf_matrix)
print(' - Accuracy Score -', accuracy_score(classi_target, predict))
print(' - Report -'), print(classification_report(classi_target, predict))
FP = cnf_matrix.sum(axis=0) - np.diag(cnf_matrix)
FN = cnf_matrix.sum(axis=1) - np.diag(cnf_matrix)
TP = np.diag(cnf_matrix)
TN = cnf_matrix.sum() - (FP + FN + TP)
FP = FP.astype(float)
print ("\nAdaBoost for Ensemble - Train Confusion Matrix\n\n",pd.crosstab(y_train,clf4_adabst_fit.predict(x_train),rownames = ["Actuall"],colnames = ["Predicted"]))
print ("\nAdaBoost for Ensemble - Train accuracy",round(accuracy_score(y_train,clf4_adabst_fit.predict(x_train)),3))
print ("\nAdaBoost for Ensemble - Train Classification Report\n",classification_report(y_train,clf4_adabst_fit.predict(x_train)))
print ("\n\nAdaBoost for Ensemble - Test Confusion Matrix\n\n",pd.crosstab(y_test,clf4_adabst_fit.predict(x_test),rownames = ["Actuall"],colnames = ["Predicted"]))
print ("\nAdaBoost for Ensemble - Test accuracy",round(accuracy_score(y_test,clf4_adabst_fit.predict(x_test)),3))
print ("\nAdaBoost for Ensemble - Test Classification Report\n",classification_report(y_test,clf4_adabst_fit.predict(x_test)))
ensemble = pd.DataFrame()
ensemble["log_output_one"] = pd.DataFrame(clf1_logreg_fit.predict_proba(x_train))[1]
ensemble["dtr_output_one"] = pd.DataFrame(clf2_dt_fit.predict_proba(x_train))[1]
ensemble["rf_output_one"] = pd.DataFrame(clf3_rf_fit.predict_proba(x_train))[1]
ensemble["adb_output_one"] = pd.DataFrame(clf4_adabst_fit.predict_proba(x_train))[1]
ensemble = pd.concat([ensemble,pd.DataFrame(y_train).reset_index(drop = True )],axis=1)
# Fitting meta-classifier
meta_logit_fit = LogisticRegression(fit_intercept=False)
meta_logit_fit.fit(ensemble[['log_output_one','dtr_output_one','rf_output_one','adb_output_one']],ensemble['Attrition_ind'])
coefs = meta_logit_fit.coef_
print ("Co-efficients for LR, DT, RF & AB are:",coefs)
ensemble_test = pd.DataFrame()
ensemble_test["log_output_one"] = pd.DataFrame(clf1_logreg_fit.predict_proba(x_test))[1]
ensemble_test["dtr_output_one"] = pd.DataFrame(clf2_dt_fit.predict_proba(x_test))[1]
ensemble_test["rf_output_one"] = pd.DataFrame(clf3_rf_fit.predict_proba(x_test))[1]
ensemble_test["adb_output_one"] = pd.DataFrame(clf4_adabst_fit.predict_proba(x_test))[1]
test_y_knn = knn_opt.predict_proba(test_x)
knn_out = submission
knn_out['target'] = test_y_knn
knn_out['target'] = 1 - knn_out['target']
knn_out.to_csv('knn_predictions1.csv', index=False, float_format='%.4f')
ada_opt = AdaBoostClassifier(algorithm='SAMME.R',
base_estimator=None,
learning_rate=1.0,
n_estimators=200,
random_state=None)
ada_opt.fit(train_x, train_y)
test_y_ada = ada_opt.predict_proba(test_x)
ada_out = submission
ada_out['target'] = test_y_ada
ada_out['target'] = 1 - ada_out['target']
ada_out.to_csv('ada_predictions1.csv', index=False, float_format='%.4f')
gb_opt = GradientBoostingClassifier(criterion='friedman_mse',
init=None,
learning_rate=0.1,
loss='deviance',
max_depth=3,
max_features=None,
max_leaf_nodes=None,
min_impurity_split=None,
def buildModel(X, y):
# X = np.reshape(X,(X.shape[0],X.shape[1] * X.shape[2]))
print X.shape, y.shape
scaler = StandardScaler()
print(scaler.fit(X))
scaled_train_x = scaler.transform(X)
X_train, X_test, y_train, y_test = train_test_split(scaled_train_x,
y,
random_state=19,
test_size=0.3)
bag = BalancedBaggingClassifier(n_estimators=200, random_state=19)
svm = SVC(class_weight='balanced',
random_state=19,
decision_function_shape='ovr')
neural = MLPClassifier(max_iter=500,
random_state=19,
solver='lbfgs',
alpha=1e-5,
hidden_layer_sizes=(49, 8, 4))
ada = AdaBoostClassifier(n_estimators=100, random_state=19)
logistic = LogisticRegression(solver='lbfgs', max_iter=500)
bag.fit(X_train, y_train)
svm.fit(X_train, y_train)
neural.fit(X_train, y_train)
ada.fit(X_train, y_train)
logistic.fit(X_train, y_train)
# joblib.dump(bag,'bag.pkl')
# joblib.dump(scaler,'scaler.pkl')
y_pred = bag.predict(X_test)
y_pred2 = svm.predict(X_test)
y_pred3 = neural.predict(X_test)
y_pred4 = ada.predict(X_test)
y_pred5 = logistic.predict(X_test)
print matthews_corrcoef(y_test, y_pred)
print matthews_corrcoef(y_test, y_pred2)
print matthews_corrcoef(y_test, y_pred3)
print matthews_corrcoef(y_test, y_pred4)
print matthews_corrcoef(y_test, y_pred5)
print confusion_matrix(y_test, y_pred)
print confusion_matrix(y_test, y_pred2)
print confusion_matrix(y_test, y_pred3)
print confusion_matrix(y_test, y_pred4)
print confusion_matrix(y_test, y_pred5)
print(classification_report_imbalanced(y_test, y_pred))
print(classification_report_imbalanced(y_test, y_pred2))
print(classification_report_imbalanced(y_test, y_pred3))
print(classification_report_imbalanced(y_test, y_pred4))
print(classification_report_imbalanced(y_test, y_pred5))
probs_ada = ada.predict_proba(X_test)
probs_bag = bag.predict_proba(X_test)
probs_neural = neural.predict_proba(X_test)
probs_logistic = logistic.predict_proba(X_test)
probs_svm = svm.decision_function(X_test)
ROCplot(probs_ada, y_test, "Plots/ROCplotADA-organelle.png")
ROCplot(probs_logistic, y_test, "Plots/ROCplotLogistic-organelle.png")
ROCplot(probs_bag, y_test, "Plots/ROCplotBAG-organelle.png")
ROCplot(probs_neural, y_test, "Plots/ROCplotNeural-organelle.png")
ROCplot(probs_svm, y_test, "Plots/ROCplotSVM-organelle.png")
multiROCplot(
[probs_ada, probs_logistic, probs_bag, probs_neural, probs_svm],
y_test, "Plots/multiROCplot.png",
['AdaBoost', 'Logistic', 'Bagging Classifier', 'MLP', 'SVM'])
#dataset is imbalanced, we'll be using the ROC AUC score as a metric instead of accuracy.
# Import DecisionTreeClassifier
from sklearn.tree import DecisionTreeClassifier
# Import AdaBoostClassifier
from sklearn.ensemble import AdaBoostClassifier
# Instantiate dt
dt = DecisionTreeClassifier(max_depth=2, random_state=1)
# Instantiate ada
ada = AdaBoostClassifier(base_estimator=dt, n_estimators=180, random_state=1)
# Fit ada to the training set
ada.fit(X_train, y_train)
# Compute the probabilities of obtaining the positive class
y_pred_proba = ada.predict_proba(X_test)[:,1]
# Import roc_auc_score
from sklearn.metrics import roc_auc_score
# Evaluate test-set roc_auc_score
ada_roc_auc = roc_auc_score(y_test, y_pred_proba)
# Print roc_auc_score
print('ROC AUC score: {:.2f}'.format(ada_roc_auc))
Clf.fit(X_train, y_train)
pred = Clf.predict_proba(X_test)[:,1]
pd.DataFrame({"id": original_test["id"], "target": pred}).to_csv("RandomForest_submission.csv", index=False)
score=cross_validate(adaClf, X_train, y_train, cv=3, scoring="roc_auc")["test_score"].mean()
print(f"{score:.6f}")
"""## AdaBoost Classifier"""
adaClf = AdaBoostClassifier()
adaClf.fit(X_train, y_train)
pred = adaClf.predict_proba(X_test)[:,1]
pd.DataFrame({"id": original_test["id"], "target": pred}).to_csv("adaboost_submission.csv", index=False)
score=cross_validate(adaClf, X_train, y_train, cv=3, scoring="roc_auc")["test_score"].mean()
print(f"{score:.6f}")
"""## GaussianProcessClassifier"""
clf = GaussianProcessClassifier(1**2 * RBF(length_scale=0.8))
clf.fit(X_train, y_train)
pred=clf.predict_proba(X_test)[:,1]
pd.DataFrame({"id": original_test["id"], "target": pred}).to_csv("GaussianProcess_submission.csv", index=False)
