gluonts.model.lstnet package

class gluonts.model.lstnet.LSTNetEstimator(freq: str, prediction_length: int, context_length: int, num_series: int, skip_size: int, ar_window: int, channels: int, lead_time: int = 0, kernel_size: int = 6, trainer: =, batch_size=None, callbacks=None, clip_gradient=10.0, ctx=None, epochs=100, hybridize=True, init="xavier", learning_rate=0.001, learning_rate_decay_factor=0.5, minimum_learning_rate=5e-05, num_batches_per_epoch=50, patience=10, weight_decay=1e-08), dropout_rate: Optional[float] = 0.2, output_activation: Optional[str] = None, rnn_cell_type: str = 'gru', rnn_num_cells: int = 100, rnn_num_layers: int = 3, skip_rnn_cell_type: str = 'gru', skip_rnn_num_layers: int = 1, skip_rnn_num_cells: int = 10, scaling: bool = True, train_sampler: Optional[gluonts.transform.sampler.InstanceSampler] = None, validation_sampler: Optional[gluonts.transform.sampler.InstanceSampler] = None, batch_size: int = 32, dtype: gluonts.core.component.DType = <class 'numpy.float32'>)[source]


Constructs an LSTNet estimator for multivariate time-series data.

The model has been described in this paper:

Note that this implementation will change over time as we further work on this method.

  • freq – Frequency of the data to train and predict

  • prediction_length – Length of the prediction p where given (y_1, …, y_t) the model predicts (y_{t+l+1}, …, y_{t+l+p}), where l is lead_time

  • context_length – The maximum number of steps to unroll the RNN for computing the predictions (Note that it is constraints by the Conv2D output size)

  • num_series – Number of time-series (covariates)

  • skip_size – Skip size for the skip RNN layer

  • ar_window – Auto-regressive window size for the linear part

  • channels – Number of channels for first layer Conv2D

  • lead_time – Lead time (default: 0)

  • kernel_size – Kernel size for first layer Conv2D (default: 6)

  • trainer – Trainer object to be used (default: Trainer())

  • dropout_rate – Dropout regularization parameter (default: 0.2)

  • output_activation – The last activation to be used for output. Accepts either None (default no activation), sigmoid or tanh

  • rnn_cell_type – Type of the RNN cell. Either lstm or gru (default: gru)

  • rnn_num_layers – Number of RNN layers to be used

  • rnn_num_cells – Number of RNN cells for each layer (default: 100)

  • skip_rnn_cell_type – Type of the RNN cell for the skip layer. Either lstm or gru ( default: gru)

  • skip_rnn_num_layers – Number of RNN layers to be used for skip part

  • skip_rnn_num_cells – Number of RNN cells for each layer for skip part (default: 10)

  • scaling – Whether to automatically scale the target values (default: True)

  • train_sampler – Controls the sampling of windows during training.

  • validation_sampler – Controls the sampling of windows during validation.

  • batch_size – The size of the batches to be used training and prediction.

  • dtype – Data type (default: np.float32)

create_predictor(transformation: gluonts.transform._base.Transformation, trained_network: mxnet.gluon.block.HybridBlock) → gluonts.model.predictor.Predictor[source]

Create and return a predictor object.


A predictor wrapping a HybridBlock used for inference.

Return type


create_training_data_loader(data: Iterable[Dict[str, Any]], **kwargs) → gluonts.dataset.loader.DataLoader[source]
create_training_network() → mxnet.gluon.block.HybridBlock[source]

Create and return the network used for training (i.e., computing the loss).


The network that computes the loss given input data.

Return type


create_transformation() → gluonts.transform._base.Transformation[source]

Create and return the transformation needed for training and inference.


The transformation that will be applied entry-wise to datasets, at training and inference time.

Return type


create_validation_data_loader(data: Iterable[Dict[str, Any]], **kwargs) → gluonts.dataset.loader.DataLoader[source]
freq = None
lead_time = None
prediction_length = None