def main():
data_train_file_path = 'Data/train.csv'
data_test_file_path = 'Data/test_with_label.csv'
data_types = {
'Survived': 'Categorical',
'Pclass': 'Categorical',
'Name_Affiliation': 'Categorical',
'Last_Name': 'Text',
'First_Name': 'Text',
'Sex': 'Categorical',
'Age': 'Numerical',
'SibSp': 'Numerical',
'Parch': 'Numerical',
'Ticket_Code': 'Categorical',
'Ticket_Number': 'Numerical',
'Fare': 'Numerical',
'Cabin_Floor': 'Categorical',
'Cabin_Rooms': 'Text',
'Embarked': 'Categorical',
'Family_Members': 'Numerical',
'is_Traveling_Alone': 'Categorical',
'Fare_Per_Passenger': 'Categorical',
'is_Female_with_Children': 'Categorical'
}
print('Loading train data...')
train_data_df = DataPreProcessor.load_data(data_train_file_path)
print('Loading test data...')
test_data_df = DataPreProcessor.load_data(data_test_file_path)
print('Cleaning and transforming train data...')
cleaned_train_data_df = DataPreProcessor.clean_data(
train_data_df, data_types)
print('Cleaning and transforming test data...')
cleaned_test_data_df = DataPreProcessor.clean_data(test_data_df,
data_types)
print('Saving cleaned train data to file...')
cleaned_train_data_df.to_csv("clean_train.csv", index=False)
print('Saving cleaned test data to file...')
cleaned_test_data_df.to_csv("clean_test.csv", index=False)
eval_classifiers = {
'TreeClassifier':
DecisionTreeClassifier(criterion="gini",
splitter="best",
max_depth=4,
random_state=42),
'AdaBoost':
AdaBoostClassifier(base_estimator=DecisionTreeClassifier(),
random_state=42),
'LogisticRegression':
LogisticRegression(penalty='l1', max_iter=10000, random_state=42),
'RandomForestClassifier':
RandomForestClassifier(n_estimators=2000,
max_depth=4,
min_samples_split=10,
random_state=42,
class_weight={
0: 0.80,
1: 0.20
}),
'GBTrees':
GradientBoostingClassifier(max_depth=4,
learning_rate=0.1,
n_estimators=2000,
random_state=42,
min_samples_split=10),
'xgboost':
XGBClassifier(max_depth=4,
n_estimators=1000,
random_state=42,
learning_rate=0.1,
min_samples_split=10),
'KNN':
KNeighborsClassifier(n_neighbors=3, p=2),
'SVM':
SVC(gamma='auto', tol=1e-3, C=1.5, random_state=42),
'GBC':
GradientBoostingClassifier(n_estimators=1000,
max_depth=4,
learning_rate=0.1)
}
eval_classifiers_params_grid = {
'TreeClassifier': {
'max_depth': [4, 5, 6]
},
'AdaBoost': {
'n_estimators': [100, 200, 500, 1000, 2000],
'learning_rate': [0.2, 0.1, 0.05, 0.01]
},
'LogisticRegression': {
'penalty': ['l1', 'l2']
},
'RandomForestClassifier': {
'n_estimators': [100, 200, 500, 1000, 2000],
'max_depth': [4, 5, 6],
'max_features': [0.8, 0.5, 0.2, 0.1]
},
'GBTrees': {
'n_estimators': [100, 500, 1000, 2000],
'max_depth': [4, 5, 6],
'max_features': [0.8, 0.5, 0.2, 0.1],
'learning_rate': [0.2, 0.1, 0.05, 0.01]
},
'xgboost': {
'n_estimators': [100, 500, 1000, 2000],
'max_depth': [4, 5, 6],
'max_features': [0.8, 0.5, 0.2, 0.1],
'learning_rate': [0.2, 0.1, 0.05, 0.01]
},
'KNN': {
'n_neighbors': [2, 3, 4, 5]
},
'SVM': {
'gamma': [0.001, 0.01, 0.1, 1],
'C': [1, 10, 50, 100, 200]
},
'GBC': {
'n_estimators': [100, 500, 1000, 2000],
'max_depth': [4, 5, 6, 8],
'max_features': [0.8, 0.5, 0.2, 0.1],
'learning_rate': [0.2, 0.1, 0.05, 0.01]
}
}
# features_cols = ['Pclass', 'Sex', 'Age_Intevals', 'Family_Members', 'Fare_Per_Passenger', 'Embarked', 'Name_Affiliation', 'Ticket_Code', 'Cabin_Floor']
# features_cols = ['Pclass', 'Sex', 'Age_Intevals', 'is_Traveling_Alone', 'Fare_Per_Passenger', 'Embarked', 'Name_Affiliation', 'Cabin_Floor']
features_cols = [
'Pclass', 'Sex', 'Age_Intervals', 'Name_Affiliation', 'Cabin_Floor',
'is_Female_with_Children'
]
one_hot_encoding_features = ['Name_Affiliation', 'Cabin_Floor', 'Pclass']
train_X, train_y = prepare_data(
cleaned_train_data_df,
class_col='Survived',
features_cols=features_cols,
one_hot_encoding_features=one_hot_encoding_features)
test_X, test_y = prepare_data(
cleaned_test_data_df,
class_col='Survived',
features_cols=features_cols,
one_hot_encoding_features=one_hot_encoding_features)
evaluator = Evaluator(train_X, train_y, test_X, test_y, eval_classifiers,
eval_classifiers_params_grid)
# evaluator.select_features(selection_clf=ExtraTreesClassifier(n_estimators=1000, max_depth=4, random_state=42))
all_predictions, final_prediction = evaluator.build_models(
grid_search=False)
evaluation_df = evaluator.save_predictions_to_df(all_predictions,
final_prediction)
submission_df = evaluator.save_predictions_for_submission(evaluation_df)
evaluation_df.to_csv("test_evaluation_results.csv", index=False)
submission_df.to_csv("test_submission.csv", index=False)
accuracy = evaluator.evaluate_performance(test_y,
final_prediction,
performance_metric='accuracy')
print('Accuracy for ensemble models {} is: {}'.format(
eval_classifiers.keys(), accuracy))
