"""
Generates a tex file containing the Latex table setup with all the relevant results.
"""
import input_data as data
import rene.analyse_results as rene_ar
import rouen.analyse_results as rouen_ar
import sherbrooke.analyse_results as sherb_ar
import stmarc.analyse_results as stmarc_ar
global_result_path_name = 'global_result/experimental_results_2.tex'
data.make_dir_if_new(global_result_path_name)
# Global result
with open(global_result_path_name, 'w') as global_result_file:
global_result_file.write(r'\documentclass{article}' + '\n\n')
global_result_file.write(r'\usepackage{multirow}' + '\n')
global_result_file.write(r'\begin{document}' + '\n\n')
global_result_file.write(r'\begin{table}' + '\n')
global_result_file.write(r'\centering' + '\n')
global_result_file.write(r'\caption{Results obtained by applying the trained model on the corresponding samples.}' \
r'\label{tab2}' + '\n')
global_result_file.write(r'\begin{tabular}{c|c|c|c||c|c|c|c|c|c|c|c}' +
'\n')
global_result_file.write(
r'\multicolumn{4}{c}{} & \multicolumn{8}{c}{Method (\%)} \\ \cline{5-12}'
+ '\n')
global_result_file.write(r'\multicolumn{4}{c}{} & \multicolumn{2}{|c|}{OC-SVM} & \multicolumn{2}{c|}{IF} & ' \
# Generate abnormal data
abnormal_data = adg.generate_abnormal_data(n_objects=20,
generate_graph=False,
show_graph=False)
abnormal_data = abnormal_data[:, 1:]
# Files setup
test_score_filename = 'results/one_class_svm/test_scores.csv'
summary_results_filename = test_score_filename[:test_score_filename.rfind(
'/')] + '/summary_results.csv'
global_summary_filename = test_score_filename[:test_score_filename.rfind(
'/')] + '/global_summary.log'
model_files_dir_name = 'model/one_class_svm/'
data.make_dir_if_new(test_score_filename)
data.make_dir_if_new(model_files_dir_name)
normal_train_ratio_list = []
normal_valid_ratio_list = []
abnormal_ratio_list = []
for i in range(aeu.repeat_number):
print('======================== Iteration {} ========================'.
format(i))
# Shuffle the data by row only
# and get the seed in order to reproduce the random sequence
train_data, validation_data, random_shuffle_seed = data.split_dataset_uniformly(
dataset)
# The trained model will be saved
abnormal_data_2 = np.genfromtxt(abnormal_file_path_2, delimiter=',')
input_size_2 = len(normal_data_2[0, 1:])
best_layer_type_2 = (256, 128, 64, 32, 16, 8)
# Files containing info of the model and threshold value
model_name_2 = 'new_method_v4_5_3'
trained_model_path_2 = 'model/' + model_name_2 + '/'
result_path_2 = 'results/' + model_name_2 + '/'
global_result_name = 'ablation_study'
result_path = 'results/' + global_result_name + '/'
global_summary_filename = result_path + 'global_ablation_study_2.csv'
data.make_dir_if_new(global_summary_filename)
with open(os.path.join(dir_name, global_summary_filename), 'wb') as summary_file:
summary_file.write(b'threshold,best_ADA_1,best_NDA_1,best_ae_ADA_1,best_ae_NDA_1,'
b'best_ADA_2,best_NDA_2,best_ae_ADA_2,best_ae_NDA_2\n')
best_ADA_list_1 = []
best_NDA_list_1 = []
best_ae_ADA_list_1 = []
best_ae_NDA_list_1 = []
best_ADA_list_2 = []
best_NDA_list_2 = []
best_ae_ADA_list_2 = []
best_ae_NDA_list_2 = []