def bids_work(fit, validate, tune=False):
print "Fit/predict bids"
model = Model()
X, y, bidder_ids = bids_to_features(fit)
model.standard_prepare(X, y, fit, validate)
clf1 = linear_model.LogisticRegression(penalty='l2', C=100.0)
clf2 = RandomForestClassifier(n_estimators=20)
clf3 = svm.SVC(probability=True, kernel='rbf', C=1, gamma=0.0001)
clf4 = GradientBoostingClassifier(loss='exponential')
clf5 = linear_model.ElasticNet(alpha=0.1, l1_ratio=0.7)
clf6 = KNeighborsClassifier()
clf7 = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
algorithm="SAMME.R",
n_estimators=2000,
learning_rate=0.01)
eclf = EnsembleClassifier(clfs=[clf1, clf2], voting='soft')
cv_fit(model.X_scaled, model.y_scaled, fit, validate,
[clf1, clf2, clf3, clf4, clf7, eclf], [
'LogisticRegression', 'RandomForest', 'SVC', 'GradientBoosting',
'AdaBoost', 'EnsembleClassifier'
], tune)
return model, eclf
def test_EnsembleClassifier_weights():
np.random.seed(123)
clf1 = LogisticRegression()
clf2 = RandomForestClassifier()
clf3 = GaussianNB()
eclf = EnsembleClassifier(clfs=[clf1, clf2, clf3], voting='soft', weights=[1,2,10])
scores = cross_validation.cross_val_score(eclf, X, y, cv=5, scoring='accuracy')
scores_mean = (round(scores.mean(), 2))
assert(scores_mean == 0.93)
def test_EnsembleClassifier_gridsearch_enumerate_names():
clf1 = LogisticRegression(random_state=1)
clf2 = RandomForestClassifier(random_state=1)
eclf = EnsembleClassifier(clfs=[clf1, clf1, clf2], voting='soft')
params = {
'logisticregression-1__C': [1.0, 100.0],
'logisticregression-2__C': [1.0, 100.0],
'randomforestclassifier__n_estimators': [5, 20],
}
grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)
gs = grid.fit(iris.data, iris.target)
def test_EnsembleClassifier_gridsearch():
clf1 = LogisticRegression(random_state=1)
clf2 = RandomForestClassifier(random_state=1)
clf3 = GaussianNB()
eclf = EnsembleClassifier(clfs=[clf1, clf2, clf3], voting='soft')
params = {
'logisticregression__C': [1.0, 100.0],
'randomforestclassifier__n_estimators': [20, 200],
}
grid = GridSearchCV(estimator=eclf, param_grid=params, cv=5)
grid.fit(iris.data, iris.target)
mean_scores = []
for params, mean_score, scores in grid.grid_scores_:
mean_scores.append(round(mean_score, 2))
assert (mean_scores == [0.95, 0.96, 0.96, 0.95])
def work(fit, validate, tune=False):
print "Fit/predict bidders"
model = Model()
X, y, ids = bidder_to_features(fit)
model.prepare(X, y, ids, fit, validate)
clf1 = linear_model.LogisticRegression(penalty='l2', C=100.0)
clf2 = RandomForestClassifier(n_estimators=20)
clf3 = svm.SVC(probability=True, kernel='rbf', C=1, gamma=0.0001)
clf4 = GradientBoostingClassifier(loss='exponential',
subsample=1,
max_features='log2',
learning_rate=0.1)
clf5 = linear_model.ElasticNet(alpha=0.1, l1_ratio=0.7)
clf6 = BaggingClassifier(n_jobs=7)
clf7 = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
algorithm="SAMME.R",
n_estimators=2000,
learning_rate=0.01)
eclf = EnsembleClassifier(clfs=[clf1, clf2], voting='soft')
# bidder_sets = split_bidders()
# for bs in bidder_sets:
# model.apply_set(bs)
# cv_fit(model.uX_scaled, model.uy_scaled, fit, validate, clf1, clf2, clf3, clf4, eclf)
cv_fit(model.X_scaled, model.y_scaled, fit, validate,
[clf1, clf2, clf3, clf4, clf7, eclf], [
'LogisticRegression', 'RandomForest', 'SVC', 'GradientBoosting',
'AdaBoost', 'EnsembleClassifier'
], tune)
#cv_fit(model.X_scaled, model.y_scaled, fit, validate, [eclf],
# ['BaggingClassifier'], tune)
return model, eclf
X_train, X_test, y_train, y_test = cross_validation.train_test_split(
vectors, labels, test_size=0.2, random_state=0)
svm = joblib.load("../classifier/svm.classifier")
lg = joblib.load("../classifier/lg.classifier")
gnb = joblib.load("../classifier/naive_bayes.classifier")
df = pandas.DataFrame(columns=('w1', 'w2', 'w3', 'mean', 'std'))
i = 0
for w1 in range(1, 4):
for w2 in range(1, 4):
for w3 in range(1, 4):
if len(set((w1, w2, w3))) == 1:
continue
eclf = EnsembleClassifier(clfs=[svm, lg, gnb],
voting='soft',
weights=[w1, w2, w3])
scores = cross_validation.cross_val_score(estimator=eclf,
X=X_test,
y=y_test,
cv=5,
scoring='accuracy',
n_jobs=1)
df.loc[i] = [w1, w2, w3, scores.mean(), scores.std()]
i += 1
df.sort(columns=['mean', 'std'], ascending=False)
print df
print "Preparing data"
vectors = []
labels = []
for row in trainingData.data:
vectors.append(row[2])
labels.append(row[1])
X_train, X_test, y_train, y_test = cross_validation.train_test_split(
vectors, labels, test_size=0.2, random_state=0)
svm = joblib.load("../classifier/svm.classifier")
lg = joblib.load("../classifier/lg.classifier")
gnb = joblib.load("../classifier/naive_bayes.classifier")
eclf = EnsembleClassifier(clfs=[svm, lg, gnb], voting='soft')
for clf, label in zip(
[svm, lg, gnb, eclf],
['SVM', 'Logistic Regression', 'Naive Bayes', 'Ensemble']):
scores = cross_validation.cross_val_score(clf,
X_test,
y_test,
cv=5,
scoring='accuracy')
print("Accuracy: %0.2f (+/- %0.2f) [%s]" %
(scores.mean(), scores.std(), label))
print classification_report(y_test, eclf.predict(X_test))