def separate_label_and_data():
util.print_debug_msg('separating label and data')
y = list(train['ACTION'])
# creating a temp object of train, so as to preserve its originality.
# i an not sure if we really need to do this.
X = train
del X['ACTION']
return X,y
def group_data(data, degree=3, hash=hash):
util.print_debug_msg('Grouping Data')
new_data = []
m,n = data.shape
#print m,n
for indicies in combinations(range(n), degree):
new_data.append([hash(tuple(v)) for v in data[:,indicies]])
return array(new_data).T
def grid_search(X_train, y_train):
util.print_debug_msg('Starting grid search')
parameters = {'C':[0.5,1,1.5,2,2.5,3,3.5,4],'penalty':['l1','l2']}
clf = GridSearchCV(lg, parameters, scoring="roc_auc", n_jobs=2)
util.print_debug_msg('Now fitting in grid search')
clf.fit(X_train, y_train)
print 'cv_scores: ', clf.cv_scores_
print 'best_estimator: ', clf.best_estimator_
print 'best_params: ', clf.best_params_
def normalize_features(X,y=None):
util.print_debug_msg('Selected Features. Now One Hot Encoding')
enc = OneHotEncoder()
if (y!=None):
enc.fit(np.vstack((X,y)))
X = enc.transform(X)
y = enc.transform(y)
return X,y
return enc.fit_transform(X)
def pre_process_data(X):
util.print_debug_msg('Pre processing Data')
x1 = array((X.ix[:]))
x2 = group_data(x1, degree=2)
x3 = group_data(x1, degree=3)
#print x1.shape
#print x2.shape
#print x3.shape
x_all = np.hstack((x1,x2,x3))
enc = OneHotEncoder()
enc.fit(x_all)
x_transformed = enc.transform(x_all).toarray()
return x_transformed
def pre_process_data(X):
util.print_debug_msg('Pre processing Data')
x1 = array((X.ix[:]))
x2 = group_data(x1, degree=2)
x3 = group_data(x1, degree=3)
#print x1.shape
#print x2.shape
#print x3.shape
x_all = np.hstack((x1,x2,x3))
#features = [0,8,9,10,36,37,38,41,41,43,47,53,60,61,63,64,67,69,71,75,85]
#features = [0, 8, 9, 10, 19, 34, 36, 37, 38, 41, 42, 43, 47, 53, 55, 60, 61, 63, 64, 67, 69, 71, 75, 81, 82, 85]
x_selected = x_all#[:,features]
'''util.print_debug_msg('Selected Features. Now One Hot Encoding')
enc = OneHotEncoder()
#enc.fit(x_selected)
#x_transformed = enc.transform(x_all)
#enc = MinMaxScaler(feature_range=(0,1), copy=False)
x_transformed = enc.fit_transform(X_selected)'''
return x_selected
def train_predict(X_train, y_train, X_test):
util.print_debug_msg('Training SVM Classifier')
model = svm.SVR(kernel='linear', verbose=True)
model.fit(X_train, y_train)
util.print_debug_msg('Predicting SVM Classifier')
return model.predict(X_test)
def train_predict(X_train, y_train, X_test):
util.print_debug_msg('Training NB Classifier')
nb = BernoulliNB()
nb.fit(X_train, y_train)
util.print_debug_msg('Predicting NB Classifier')
return nb.predict(X_test)
def train_predict(X_train, y_train, X_test):
util.print_debug_msg('Training kNN Classifier')
knn = KNeighborsClassifier(n_neighbors=10)
knn.fit(X_train, y_train)
util.print_debug_msg('Predicting kNN Classifier')
return knn.predict_proba(X_test)[:, 1]
def train_predict(X_train, y_train, X_test):
util.print_debug_msg('Training LG Classifier')
lg.fit(X_train, y_train)
util.print_debug_msg('Predicting LG Classifier')
return lg.predict_proba(X_test)[:, 1]
