from sklearn.model_selection import train_test_split from sklearn.model_selection import GridSearchCV from sklearn.linear_model import LogisticRegression from sklearn.ensemble import RandomForestClassifier, VotingClassifier import sklearn.svm as svm from sklearn.utils import shuffle from sklearn.externals import joblib from sklearn.metrics import accuracy_score, f1_score from sklearn.model_selection import cross_val_score import numpy as np import pandas as pd import sys, os, getopt saved_models_folder = 'saved_models' current_dirpath = os.getcwd() output_model_folder = os.path.join(current_dirpath, saved_models_folder) def get_best_model(X_train, y_train): Cs = [0.001, 0.01, 0.1, 1, 10, 100, 1000] gammas = [0.001, 0.01, 0.1, 1, 5, 10, 100] param_grid = {'kernel':['rbf'], 'C': Cs, 'gamma' : gammas} svc = svm.SVC(probability=True) clf = GridSearchCV(svc, param_grid, cv=10, scoring='accuracy', verbose=10) clf.fit(X_train, y_train) model = clf.best_estimator_ return model def main(): if len(sys.argv) <= 1: print('Run with default parameters...') print('python ensemble_model_train.py --data xxxx --output xxxx') sys.exit(2) try: opts, args = getopt.getopt(sys.argv[1:], "hd:o", ["help=", "data=", "output="]) except getopt.GetoptError: # print help information and exit: print('python ensemble_model_train.py --data xxxx --output xxxx') sys.exit(2) for o, a in opts: if o == "-h": print('python ensemble_model_train.py --data xxxx --output xxxx') sys.exit() elif o in ("-d", "--data"): p_data_file = a elif o in ("-o", "--output"): p_output = a else: assert False, "unhandled option" if not os.path.exists(output_model_folder): os.makedirs(output_model_folder) ######################## # 1. Get and prepare data ######################## dataset_train = pd.read_csv(p_data_file + '.train', header=None, sep=";") dataset_test = pd.read_csv(p_data_file + '.test', header=None, sep=";") # default first shuffle of data dataset_train = shuffle(dataset_train) dataset_test = shuffle(dataset_test) # get dataset with equal number of classes occurences noisy_df_train = dataset_train[dataset_train.ix[:, 0] == 1] not_noisy_df_train = dataset_train[dataset_train.ix[:, 0] == 0] nb_noisy_train = len(noisy_df_train.index) noisy_df_test = dataset_test[dataset_test.ix[:, 0] == 1] not_noisy_df_test = dataset_test[dataset_test.ix[:, 0] == 0] nb_noisy_test = len(noisy_df_test.index) final_df_train = pd.concat([not_noisy_df_train[0:nb_noisy_train], noisy_df_train]) final_df_test = pd.concat([not_noisy_df_test[0:nb_noisy_test], noisy_df_test]) # shuffle data another time final_df_train = shuffle(final_df_train) final_df_test = shuffle(final_df_test) final_df_train_size = len(final_df_train.index) final_df_test_size = len(final_df_test.index) # use of the whole data set for training x_dataset_train = final_df_train.ix[:,1:] x_dataset_test = final_df_test.ix[:,1:] y_dataset_train = final_df_train.ix[:,0] y_dataset_test = final_df_test.ix[:,0] ####################### # 2. Construction of the model : Ensemble model structure ####################### svm_model = get_best_model(x_dataset_train, y_dataset_train) lr_model = LogisticRegression(solver='liblinear', multi_class='ovr', random_state=1) rf_model = RandomForestClassifier(n_estimators=100, random_state=1) ensemble_model = VotingClassifier(estimators=[ ('svm', svm_model), ('lr', lr_model), ('rf', rf_model)], voting='soft', weights=[1,1,1]) ####################### # 3. Fit model : use of cross validation to fit model ####################### print("-------------------------------------------") print("Train dataset size: ", final_df_train_size) ensemble_model.fit(x_dataset_train, y_dataset_train) val_scores = cross_val_score(ensemble_model, x_dataset_train, y_dataset_train, cv=5) print("Accuracy: %0.2f (+/- %0.2f)" % (val_scores.mean(), val_scores.std() * 2)) ###################### # 4. Test : Validation and test dataset from .test dataset ###################### # we need to specify validation size to 20% of whole dataset val_set_size = int(final_df_train_size/3) test_set_size = val_set_size total_validation_size = val_set_size + test_set_size if final_df_test_size > total_validation_size: x_dataset_test = x_dataset_test[0:total_validation_size] y_dataset_test = y_dataset_test[0:total_validation_size] X_test, X_val, y_test, y_val = train_test_split(x_dataset_test, y_dataset_test, test_size=0.5, random_state=1) y_test_model = ensemble_model.predict(X_test) y_val_model = ensemble_model.predict(X_val) val_accuracy = accuracy_score(y_val, y_val_model) test_accuracy = accuracy_score(y_test, y_test_model) val_f1 = f1_score(y_val, y_val_model) test_f1 = f1_score(y_test, y_test_model) ################### # 5. Output : Print and write all information in csv ################### print("Validation dataset size ", val_set_size) print("Validation: ", val_accuracy) print("Validation F1: ", val_f1) print("Test dataset size ", test_set_size) print("Test: ", val_accuracy) print("Test F1: ", test_f1) ################## # 6. Save model : create path if not exists ################## if not os.path.exists(saved_models_folder): os.makedirs(saved_models_folder) joblib.dump(ensemble_model, output_model_folder + '/' + p_output + '.joblib') if __name__== "__main__": main()