def custom2(param, comb_degree=3):
print(f"{dt()} :: Experiments Initialize")
for nb_combine in range(1, comb_degree+1):
print(f"{dt()} :: {nb_combine} sample experiments")
param.nb_combine = nb_combine
datasets = loader.data_loader(param, target=nb_combine)
datasets = preprocessing.del_subject(param, datasets, target="PA")
train, test, nb_class, nb_people = preprocessing.chosen_method(param=param, comb=nb_combine, datasets=datasets)
deep_learning_experiment_configuration(param, train, test, [nb_class, nb_people])
ds.save_result(param)
def experiment(param, comb_degree=5):
print(f"{dt()} :: Experiments Initialize")
for nb_combine in [2, 4]:
# if nb_combine != 1:
# continue
print(f"{dt()} :: {nb_combine} sample experiments")
param.nb_combine = nb_combine
if param.model_name in model_compactor.model_info['dl']:
datasets = loader.data_loader(param, target=nb_combine)
train, test, nb_class, nb_people = preprocessing.chosen_method(
param=param, comb=nb_combine, datasets=datasets)
deep_learning_experiment_configuration(param, train, test,
[nb_class, nb_people])
ds.save_result(param)
elif param.model_name in model_compactor.model_info['c_dl']:
datasets = loader.data_loader(param, target=nb_combine)
train, test, nb_class, nb_people = preprocessing.chosen_method(
param=param, comb=nb_combine, datasets=datasets)
deep_learning_experiment_custom(param, train, test,
[nb_class, nb_people])
ds.save_result(param)
elif param.model_name in model_compactor.model_info['v_dl']:
datasets = loader.vector_loader(param)
train, test, nb_class, nb_people = preprocessing.chosen_method(
param=param, comb=1, datasets=datasets)
deep_learning_experiment_vector(param, train, test,
[nb_class, nb_people])
ds.save_result(param)
elif param.object == 'ensemble':
datasets = loader.data_loader(param, nb_combine)
train, test, nb_class, nb_people = preprocessing.chosen_method(
param=param, comb=nb_combine, datasets=datasets)
deep_learning_experiment_ensemble(param, train, test,
[nb_class, nb_people])
ds.save_result(param)
def experiment(param, comb_degree=5):
print(f"{dt()} :: Experiments Initialize")
for nb_combine in range(1, comb_degree+1):
print(f"{dt()} :: {nb_combine} sample experiments")
param.nb_combine = nb_combine
if nb_combine != 1:
continue
datasets = loader.data_loader(param, target=nb_combine)
train, test, nb_class, nb_people = preprocessing.chosen_method(param=param, comb=nb_combine, datasets=datasets)
if param.model_name in model_compactor.model_info['dl']:
deep_learning_experiment_configuration(param, train, test, [nb_class, nb_people])
ds.save_result(param)
elif param.model_name in model_compactor.model_info['ml']:
machine_learning_experiment_configuration(param, train, test, [nb_class, nb_people])
def convert(param):
print(f"{dt()} :: Convert Initialize")
for nb_combine in range(1, 5):
print(f"{dt()} :: {nb_combine} sample experiments")
param.nb_combine = nb_combine
if nb_combine != 1:
continue
datasets = loader.data_loader(param, target=nb_combine)
train, test, nb_class, nb_people = preprocessing.chosen_method(
param=param, comb=nb_combine, datasets=datasets)
if param.method == method_select['people']:
nb_repeat = nb_people
elif param.method in method_select['repeat']:
nb_repeat = 20
elif param.method in method_select["CrossValidation"]:
nb_repeat = param.collect["CrossValidation"] * 5
elif param.method in method_select['specific']:
nb_repeat = 5
for repeat in range(nb_repeat):
print(
f"{dt()} :: {repeat + 1}/{nb_repeat} convert target progress")
tartr = train[repeat]
tarte = test[repeat]
if param.datatype == "type":
tartr["tag"] -= 1
tarte["tag"] -= 1
tr_label = np.zeros([len(tartr["people"]), 2])
te_label = np.zeros([len(tarte["people"]), 2])
for idx, (tr, te) in enumerate(
zip([tartr["people"], tartr["tag"]],
[tarte["people"], tarte["tag"]])):
tr_label[:, idx] = tr
te_label[:, idx] = te
for idx in range(param.nb_modal):
tartr[f"data_{idx}"] = np.hstack(
[tartr[f"data_{idx}"], tr_label])
tarte[f"data_{idx}"] = np.hstack(
[tarte[f"data_{idx}"], te_label])
tr_data = list()
te_data = list()
for idx in range(param.nb_modal):
tr_data.append(tartr[f"data_{idx}"])
te_data.append(tarte[f"data_{idx}"])
train_dict = dict()
test_dict = dict()
if param.nb_modal == 3:
datatype = ['pressure', 'acc', 'gyro']
for train_target, test_target, target in zip(
tr_data, te_data, datatype):
train_dict[target] = train_target
test_dict[target] = test_target
elif param.nb_modal == 7:
datatype = ['pressure', 'acc', 'gyro']
for train_target, test_target, target in zip(
tr_data, te_data, datatype):
train_dict[target] = train_target
test_dict[target] = test_target
save_dir = '../Result/Convert'
train_folder = 'train'
test_folder = 'test'
folder_name = 'matfile'
file_name = f'{repeat}.mat'
train_dir = None
test_dir = None
for idx, target in enumerate([save_dir, folder_name,
train_folder]):
if idx == 0:
train_dir = target
else:
train_dir = os.path.join(train_dir, target)
if os.path.exists(train_dir) is not True:
os.mkdir(train_dir)
for idx, target in enumerate([save_dir, folder_name, test_folder]):
if idx == 0:
test_dir = target
else:
test_dir = os.path.join(test_dir, target)
if os.path.exists(test_dir) is not True:
os.mkdir(test_dir)
savemat(os.path.join(train_dir, file_name), train_dict)
savemat(os.path.join(test_dir, file_name), test_dict)