def inspect_final_data_set_with_labels(image_index_list, seed):
np.random.seed(seed)
# Cross Validation
fold_num = 5
unique_label_values = np.unique(image_index_list)
selected_label_values = np.random.choice(unique_label_values, \
size=np.ceil(unique_label_values.size * (fold_num - 1) / fold_num), \
replace=False)
selected_index_list = []
for single_image_index in image_index_list:
if single_image_index in selected_label_values:
selected_index_list.append(single_image_index)
selected_index_array = np.array(selected_index_list)
_, Y_train = solution_basic.get_record_map(selected_index_array, None)
true_records = Y_train == 1
true_records_num = np.sum(true_records)
false_records_num = Y_train.size - true_records_num
return ([true_records_num], [false_records_num])
def inspect_final_data_set_with_labels(image_index_list, seed):
np.random.seed(seed)
# Cross Validation
fold_num = 5
unique_label_values = np.unique(image_index_list)
selected_label_values = np.random.choice(
unique_label_values, size=np.ceil(unique_label_values.size * (fold_num - 1) / fold_num), replace=False
)
selected_index_list = []
for single_image_index in image_index_list:
if single_image_index in selected_label_values:
selected_index_list.append(single_image_index)
selected_index_array = np.array(selected_index_list)
_, Y_train = solution_basic.get_record_map(selected_index_array, None)
true_records = Y_train == 1
true_records_num = np.sum(true_records)
false_records_num = Y_train.size - true_records_num
return ([true_records_num], [false_records_num])
def inspect_final_data_set_without_labels(image_index_list, seed):
np.random.seed(seed)
image_index_array = np.array(image_index_list)
# Cross Validation
fold_num = 5
label_kfold = KFold(image_index_array.size, n_folds=fold_num, shuffle=True)
true_records_num_list = []
false_records_num_list = []
for _, fold_item in enumerate(label_kfold):
# Generate final data set
selected_index_array = image_index_array[fold_item[0]]
_, Y_train = solution_basic.get_record_map(selected_index_array, None)
true_records = Y_train == 1
true_records_num = np.sum(true_records)
false_records_num = Y_train.size - true_records_num
true_records_num_list.append(true_records_num)
false_records_num_list.append(false_records_num)
return (true_records_num_list, false_records_num_list)
def inspect_final_data_set_without_labels(image_index_list, seed):
np.random.seed(seed)
image_index_array = np.array(image_index_list)
# Cross Validation
fold_num = 5
label_kfold = KFold(image_index_array.size, n_folds=fold_num, shuffle=True)
true_records_num_list = []
false_records_num_list = []
for _, fold_item in enumerate(label_kfold):
# Generate final data set
selected_index_array = image_index_array[fold_item[0]]
_, Y_train = solution_basic.get_record_map(selected_index_array, None)
true_records = Y_train == 1
true_records_num = np.sum(true_records)
false_records_num = Y_train.size - true_records_num
true_records_num_list.append(true_records_num)
false_records_num_list.append(false_records_num)
return (true_records_num_list, false_records_num_list)
