def gradient_boosting_cross(X, Y, fold):
clf = GradientBoostingClassifier(n_estimators=200,
random_state=0,
max_depth=15)
ypred = cross_val_predict(clf, X, Y, cv=fold)
print("Mean Squared Error: ", mean_squared_error(y_true=Y, y_pred=ypred))
util.plot_roc(ypred, Y)
util.confusion_matrix_report(['Default', 'Good'], ypred, Y)
return {'Gradient Boosting': ypred}
def random_forest_cross(X, Y, fold):
clf = RandomForestClassifier(n_estimators=200,
n_jobs=30,
random_state=0,
verbose=0)
ypred = cross_val_predict(clf, X, Y, cv=fold)
print("Mean Squared Error: ", mean_squared_error(y_true=Y, y_pred=ypred))
util.plot_roc(ypred, Y)
util.confusion_matrix_report(['Default', 'Good'], ypred, Y)
return {'Floresta Aleatória': ypred}
def plot_roc(self, X, y, size_x, size_y):
"""Plot the ROC curve for X_test and y_test.
"""
plot_roc(self.clf, X, y, size_x, size_y)
@author: jason
'''
from util import loaddata, plot_roc, roc
from pipeline import Pipeline
from pyearth.earth import Earth
from sklearn.linear_model.logistic import LogisticRegression
from matplotlib import pyplot
from sklearn.linear_model.stochastic_gradient import SGDClassifier
from sklearn.ensemble.weight_boosting import AdaBoostClassifier
from sklearn.base import ClassifierMixin
class ClassifierPipeline(Pipeline, ClassifierMixin):
@property
def classes_(self):
return self.steps[-1][1].classes_
wtrain, wtest, Xtrain, Xtest, ytrain, ytest = loaddata(nrows=5000)
classifier = ClassifierPipeline([('earth',Earth(max_degree=1)), ('log',LogisticRegression())])
model = classifier#AdaBoostClassifier(base_estimator=classifier)
model = model.fit(Xtrain, ytrain)
ptest = model.predict_proba(Xtest)[:,1]
print model.steps[0][1].trace()
print model.steps[0][1].summary()
pyplot.figure()
plot_roc(*roc(ytest, ptest))
pyplot.show()
# print train[train.columns[-5:-3]]
def plot_roc(self, X, y, size_x, size_y):
'''Plot the ROC curve for X_test and y_test'''
plot_roc(self.clf, X, y, size_x, size_y)
def plot_roc(self,X,y,size_x, size_y):
plot_roc(self.clf,X,y,size_x,size_y)
