def model_lda(train, test, label):
reglin = discriminant_analysis.LinearDiscriminantAnalysis()
reglin.fit(train[label], train['hotel_cluster'])
prediction = reglin.predict_proba(test[label])
return util.best_proba(prediction), reglin
def model_knn(train, test, label):
neigh = KNeighborsClassifier(n_neighbors=30)
neigh.fit(train[label], train['hotel_cluster'])
prediction = neigh.predict_proba(test[label])
return util.best_proba(prediction), neigh
def model_reglog(train, test, label):
reglog = linear_model.LogisticRegression()
reglog.fit(train[label], train['hotel_cluster'])
prediction = reglog.predict_proba(test[label])
return util.best_proba(prediction), reglog
def model_lasso(train, test, label):
C = 0.01
lasso = linear_model.LogisticRegression(C=C, penalty="l2")
lasso.fit(train[label], train['hotel_cluster'])
prediction = lasso.predict_proba(test[label])
return util.best_proba(prediction), lasso
def model_rforest(train,test,label):
rf = ske.RandomForestClassifier(n_estimators=10,criterion="gini")
rf.fit(train[label],train['hotel_cluster'])
prediction = rf.predict_proba(test[label])
return util.best_proba(prediction), rf
def model_SVM(train, test, label):
SVM = svm.SVC(kernel='rbf', probability=True)
SVM.fit(train[label], train['hotel_cluster'])
prediction = SVM.predict_proba(test[label])
return util.best_proba(prediction), SVM
def model_labelprop(train, test, label):
lp = sm.LabelPropagation(kernel='rbf')
lp.fit(train[label], train['hotel_cluster'])
prediction = lp.predict_proba(test[label])
return util.best_proba(prediction), lp
def model_adaboost(train, test, label):
adab = AdaBoostClassifier(learning_rate=0.1, n_estimators=100)
adab.fit(train[label], train['hotel_cluster'])
prediction = adab.predict_proba(test[label])
return util.best_proba(prediction), adab
def model_dec_tree(train,test,label):
dectree = DecisionTreeClassifier(criterion='gini', splitter='best', max_depth=None, max_features=None)
dectree.fit(train[label], train["hotel_cluster"])
prediction = dectree.predict_proba(test[label])
return util.best_proba(prediction), dectree
def model_gradboost(train, test, label):
gradb = GradientBoostingClassifier(n_estimators=10)
gradb.fit(train[label], train['hotel_cluster'])
prediction = gradb.predict_proba(test[label])
return util.best_proba(prediction), gradb
def model_xgb(train, test, label):
xgb = sklearn.XGBClassifier(nthread=4, n_estimators=10)
xgb.fit(train[label], train['hotel_cluster'])
prediction = xgb.predict_proba(test[label])
df = pd.DataFrame(prediction).transpose().tail(test[label].shape[0])
return util.best_proba(df.as_matrix()), xgb
def error(weight, matrix_pred, test):
prediction_final = map(np.multiply, weight, matrix_pred)
prediction_final = np.sum(prediction_final, axis=0)
clusters = util.best_proba(prediction_final)
def temp(i, val):
actual[i] = [val]
actual = range(len(test['hotel_cluster']))
map(temp, range(len(test['hotel_cluster'])), test['hotel_cluster'])
return 1 - util.mapk(actual, clusters)
def model_neural(train, test, label):
model = Sequential()
model.add(
Dense(12, input_dim=len(label), init='uniform', activation='relu'))
model.add(Dense(len(label), init='uniform', activation='relu'))
model.add(Dense(1, init='uniform', activation='sigmoid'))
model.compile(loss='binary_crossentropy',
optimizer='adam',
metrics=['accuracy'])
model.fit(train[label],
train['hotel_cluster'],
nb_epoch=150,
batch_size=10)
prediction = model.predict_proba(test[label])
return util.best_proba(prediction)
def model_bagging(train, test, label, grid_elt=None):
if grid_elt is None:
bag = ensemble.BaggingClassifier(KNeighborsClassifier(),
n_jobs=1,
n_estimators=30,
max_samples=1.0,
max_features=0.1)
else:
bag = ensemble.BaggingClassifier(KNeighborsClassifier(),
n_jobs=1,
n_estimators=grid_elt[0],
max_samples=grid_elt[1],
max_features=grid_elt[2])
bag.fit(train[label], train['hotel_cluster'])
prediction = bag.predict_proba(test[label])
return util.best_proba(prediction), bag
def model_voting(train, test, label, grid_elt=None):
clf1 = LogisticRegression(random_state=1)
clf2 = GradientBoostingClassifier(n_estimators=10)
clf3 = KNeighborsClassifier(n_neighbors=30)
if grid_elt is None:
eclf1 = VotingClassifier(estimators=[('lr', clf1), ('gb', clf2),
('knn', clf3)],
voting='soft',
weights=[1, 1, 1])
else:
eclf1 = VotingClassifier(estimators=[('lr', clf1), ('gb', clf2),
('knn', clf3)],
voting='soft',
weights=grid_elt)
eclf1.fit(train[label], train['hotel_cluster'])
prediction = eclf1.predict_proba(test[label])
return util.best_proba(prediction), eclf1
def model_weighted(train, test, label):
weight = []
reglog = linear_model.LogisticRegression()
reglog.fit(train[label], train['hotel_cluster'])
prediction_reglog = reglog.predict_proba(test[label])
gradb = GradientBoostingClassifier(n_estimators=10)
gradb.fit(train[label], train['hotel_cluster'])
prediction_gradboost = gradb.predict_proba(test[label])
neigh = KNeighborsClassifier(n_neighbors=10)
neigh.fit(train[label], train['hotel_cluster'])
prediction_knn = neigh.predict_proba(test[label])
matrix_pred = [prediction_reglog, prediction_gradboost, prediction_knn]
prediction_final = map(np.multiply, weight, matrix_pred)
prediction_final = np.sum(prediction_final, axis=0)
return util.best_proba(prediction_final)
