from sklearn.naive_bayes import BernoulliNB
from run_binary_classifier import run
param_grid = {
'bag_of_words__stop_words': ['english'],
'bag_of_words__ngram_range': [(1, 2)],
'bag_of_words__max_features': [500],
'dim_reduct__n_components': [300],
'normalizer__norm': ['l2'],
'classifier__alpha': [1.0],
'classifier__binarize': [0.0]
}
clf = BernoulliNB()
run(param_grid, clf)
def keras_logreg_model():
model = Sequential()
model.add(
Dense(units=1,
input_shape=(2, ),
kernel_initializer='normal',
kernel_regularizer=regularizers.l2(1.),
activation='sigmoid'))
model.compile(loss='binary_crossentropy',
optimizer='adam',
metrics=['accuracy'])
return model
param_grid = {
'bag_of_words__stop_words': ['english'],
'bag_of_words__ngram_range': [(1, 2)],
'bag_of_words__max_features': [500],
#'dim_reduct__n_components': [300],
'normalizer__norm': ['l2']
#'classifier__C': [5., 10.]
}
estimator = KerasClassifier(build_fn=keras_logreg_model,
epochs=1,
batch_size=5,
verbose=1)
run(param_grid, estimator)
multilabel_param_grid = [{
'estimator__bag_of_words__stop_words': ['english'],
'estimator__bag_of_words__ngram_range': [(1, 2)],
'estimator__bag_of_words__max_features': [500],
'estimator__dim_reduct__n_components': [300],
'estimator__normalizer__norm': ['l2'],
'estimator__classifier__C': [5., 10.]
}]
# =========================== #
# TRAIN
# BINARY CLASSIFIER
# =========================== #
binary_clf = run_binary_classifier.run(binary_param_grid, LogisticRegression(), comments_file=train_binary)
with open('./saved_models/log_reg_joint_binary.pkl', 'wb') as saved_model:
pickle.dump(binary_clf, file=saved_model)
# =========================== #
# TRAIN
# MULTILABEL CLASSIFIER
# =========================== #
multilabel_clf = run_multilabel_classifier.run(multilabel_param_grid, LogisticRegression(), comments_file=train_multilabel)
with open('./saved_models/log_reg_joint_multilabel.pkl', 'wb') as saved_model:
pickle.dump(binary_clf, file=saved_model)
# =========================== #
# PREDICT
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import classification_report
from run_binary_classifier import _load_comments, run
train_comments_path = os.path.join('../../../', 'data/train_binary.csv')
test_comments_path = os.path.join('../../../', 'data/test_clean_binary.csv')
param_grid = {
'bag_of_words__stop_words': ['english'],
'bag_of_words__ngram_range': [(1, 2)],
'bag_of_words__max_features': [500],
'dim_reduct__n_components': [300],
'normalizer__norm': ['l2'],
'classifier__C': [5., 10.]
}
clf = LogisticRegression()
trained_clf = run(param_grid, clf, comments_file=train_comments_path)
with open('./saved_models/log_reg_trained_binary.pkl', 'wb') as saved_model:
pickle.dump(trained_clf, file=saved_model)
with open('./saved_models/log_reg_trained_binary.pkl', 'rb') as saved_model:
loaded_clf = pickle.load(saved_model)
X_test, y_test = _load_comments(test_comments_path)
y_test_predict = loaded_clf.predict(X_test)
print(classification_report(y_test, y_test_predict))