def test_tcn_forecaster_shape_error(self):
train_data, val_data, test_data = create_data()
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=1,
loss="mae",
lr=0.01)
with pytest.raises(AssertionError):
forecaster.fit(train_data, epochs=2)
def test_tcn_forecaster_fit_eva_pred(self):
train_data, val_data, test_data = create_data()
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
hidden_dim=[32, 16],
layer_num=2,
dropout=[0.1, 0.2],
loss="mae",
lr=0.01)
train_loss = forecaster.fit(train_data, epochs=2)
test_pred = forecaster.predict(test_data[0])
assert test_pred.shape == test_data[1].shape
test_mse = forecaster.evaluate(test_data)
def test_tcn_forecaster_save_load(self):
train_data, val_data, test_data = create_data()
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
loss="mae",
lr=0.01)
train_mse = forecaster.fit(train_data, epochs=2)
with tempfile.TemporaryDirectory() as tmp_dir_name:
ckpt_name = os.path.join(tmp_dir_name, "ckpt")
test_pred_save = forecaster.predict(test_data[0])
forecaster.save(ckpt_name)
forecaster.load(ckpt_name)
test_pred_load = forecaster.predict(test_data[0])
np.testing.assert_almost_equal(test_pred_save, test_pred_load)
def test_fit_score(self):
look_back = 4
# generate dataframe
data = self.gen_data(feature_num=6, sample_num=100)
# split train and test dataframes
train_df, test_df = self.train_test_split(data, test_num=20, look_back=look_back)
# roll data to generate model input
x_train, y_train = self.roll_data(dataset=train_df, look_back=look_back,
target_col_indexes=[0])
x_test, y_test = self.roll_data(dataset=test_df, look_back=look_back,
target_col_indexes=[0])
# create model, train on train data and predict on test
y_train = np.expand_dims(y_train, 1)
forecaster = LSTMForecaster(past_seq_len=look_back,
input_feature_num=x_train.shape[-1],
output_feature_num=1,
hidden_dim=32,
layer_num=2)
forecaster.fit(data=(x_train, y_train), batch_size=1024, epochs=50)
y_predict = forecaster.predict(x_test)
y_predict = np.squeeze(y_predict, axis=1)
# find anomaly using a manual set threshold
td = ThresholdDetector()
td.set_params(threshold=10)
td.fit(y_test, y_predict)
anomaly_scores = td.score()
assert len(list(np.where(anomaly_scores > 0)[0])) == 0
anomaly_indexes = td.anomaly_indexes()
assert len(anomaly_indexes) == 0
# if threshold is not provided, ThresholdDetector can fit to the data
ratio = 0.1
td = ThresholdDetector()
td.set_params(ratio=ratio)
td.fit(y_test, y_predict)
fitted_anomaly_indexes = td.anomaly_indexes()
assert len(fitted_anomaly_indexes) == int(ratio * y_test.shape[0])
def test_tcn_forecaster_xshard_input(self):
train_data, val_data, test_data = create_data()
print("original", train_data[0].dtype)
init_orca_context(cores=4, memory="2g")
from zoo.orca.data import XShards
def transform_to_dict(data):
return {'x': data[0], 'y': data[1]}
def transform_to_dict_x(data):
return {'x': data[0]}
train_data = XShards.partition(train_data).transform_shard(
transform_to_dict)
val_data = XShards.partition(val_data).transform_shard(
transform_to_dict)
test_data = XShards.partition(test_data).transform_shard(
transform_to_dict_x)
for distributed in [True, False]:
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
loss="mae",
lr=0.01,
distributed=distributed)
forecaster.fit(train_data, epochs=2)
distributed_pred = forecaster.predict(test_data)
distributed_eval = forecaster.evaluate(val_data)
stop_orca_context()
def test_tcn_forecaster_runtime_error(self):
train_data, val_data, test_data = create_data()
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
loss="mae",
lr=0.01)
with pytest.raises(RuntimeError):
with tempfile.TemporaryDirectory() as tmp_dir_name:
ckpt_name = os.path.join(tmp_dir_name, "ckpt")
forecaster.save(ckpt_name)
with pytest.raises(RuntimeError):
forecaster.predict(test_data[0])
with pytest.raises(RuntimeError):
forecaster.evaluate(test_data)
def test_tcn_forecaster_onnx_methods(self):
train_data, val_data, test_data = create_data()
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
loss="mae",
lr=0.01)
forecaster.fit(train_data, epochs=2)
try:
import onnx
import onnxruntime
pred = forecaster.predict(test_data[0])
pred_onnx = forecaster.predict_with_onnx(test_data[0])
np.testing.assert_almost_equal(pred, pred_onnx, decimal=5)
mse = forecaster.evaluate(test_data, multioutput="raw_values")
mse_onnx = forecaster.evaluate_with_onnx(test_data,
multioutput="raw_values")
np.testing.assert_almost_equal(mse, mse_onnx, decimal=5)
with pytest.raises(RuntimeError):
forecaster.build_onnx(sess_options=1)
sess_options = onnxruntime.SessionOptions()
sess_options.intra_op_num_threads = 1
forecaster.build_onnx(sess_options=sess_options)
mse = forecaster.evaluate(test_data)
mse_onnx = forecaster.evaluate_with_onnx(test_data)
np.testing.assert_almost_equal(mse, mse_onnx, decimal=5)
except ImportError:
pass
def test_tcn_forecaster_distributed(self):
train_data, val_data, test_data = create_data()
init_orca_context(cores=4, memory="2g")
forecaster = LSTMForecaster(past_seq_len=24,
input_feature_num=2,
output_feature_num=2,
loss="mae",
lr=0.01,
distributed=True)
forecaster.fit(train_data, epochs=2)
distributed_pred = forecaster.predict(test_data[0])
distributed_eval = forecaster.evaluate(val_data)
model = forecaster.get_model()
assert isinstance(model, torch.nn.Module)
forecaster.to_local()
local_pred = forecaster.predict(test_data[0])
local_eval = forecaster.evaluate(val_data)
np.testing.assert_almost_equal(distributed_pred, local_pred, decimal=5)
try:
import onnx
import onnxruntime
local_pred_onnx = forecaster.predict_with_onnx(test_data[0])
local_eval_onnx = forecaster.evaluate_with_onnx(val_data)
np.testing.assert_almost_equal(distributed_pred,
local_pred_onnx,
decimal=5)
except ImportError:
pass
model = forecaster.get_model()
assert isinstance(model, torch.nn.Module)
stop_orca_context()