"""
Temporal Convolutional Network
------------------------------
"""
import math
from typing import Optional, Sequence, Tuple
import torch
import torch.nn as nn
import torch.nn.functional as F
from darts.logging import get_logger, raise_if_not
from darts.models.forecasting.pl_forecasting_module import (
PLPastCovariatesModule,
io_processor,
)
from darts.models.forecasting.torch_forecasting_model import PastCovariatesTorchModel
from darts.timeseries import TimeSeries
from darts.utils.data import PastCovariatesShiftedDataset
from darts.utils.torch import MonteCarloDropout
logger = get_logger(__name__)
class _ResidualBlock(nn.Module):
def __init__(
self,
num_filters: int,
kernel_size: int,
dilation_base: int,
dropout: float,
weight_norm: bool,
nr_blocks_below: int,
num_layers: int,
input_size: int,
target_size: int,
):
"""PyTorch module implementing a residual block module used in `_TCNModule`.
Parameters
----------
num_filters
The number of filters in a convolutional layer of the TCN.
kernel_size
The size of every kernel in a convolutional layer.
dilation_base
The base of the exponent that will determine the dilation on every level.
dropout
The dropout to be applied to every convolutional layer.
weight_norm
Boolean value indicating whether to use weight normalization.
nr_blocks_below
The number of residual blocks before the current one.
num_layers
The number of convolutional layers.
input_size
The dimensionality of the input time series of the whole network.
target_size
The dimensionality of the output time series of the whole network.
Inputs
------
x of shape `(batch_size, in_dimension, input_chunk_length)`
Tensor containing the features of the input sequence.
in_dimension is equal to `input_size` if this is the first residual block,
in all other cases it is equal to `num_filters`.
Outputs
-------
y of shape `(batch_size, out_dimension, input_chunk_length)`
Tensor containing the output sequence of the residual block.
out_dimension is equal to `output_size` if this is the last residual block,
in all other cases it is equal to `num_filters`.
"""
super().__init__()
self.dilation_base = dilation_base
self.kernel_size = kernel_size
self.dropout1 = MonteCarloDropout(dropout)
self.dropout2 = MonteCarloDropout(dropout)
self.num_layers = num_layers
self.nr_blocks_below = nr_blocks_below
input_dim = input_size if nr_blocks_below == 0 else num_filters
output_dim = target_size if nr_blocks_below == num_layers - 1 else num_filters
self.conv1 = nn.Conv1d(
input_dim,
num_filters,
kernel_size,
dilation=(dilation_base**nr_blocks_below),
)
self.conv2 = nn.Conv1d(
num_filters,
output_dim,
kernel_size,
dilation=(dilation_base**nr_blocks_below),
)
if weight_norm:
self.conv1, self.conv2 = nn.utils.weight_norm(
self.conv1
), nn.utils.weight_norm(self.conv2)
if input_dim != output_dim:
self.conv3 = nn.Conv1d(input_dim, output_dim, 1)
def forward(self, x):
residual = x
# first step
left_padding = (self.dilation_base**self.nr_blocks_below) * (
self.kernel_size - 1
)
x = F.pad(x, (left_padding, 0))
x = self.dropout1(F.relu(self.conv1(x)))
# second step
x = F.pad(x, (left_padding, 0))
x = self.conv2(x)
if self.nr_blocks_below < self.num_layers - 1:
x = F.relu(x)
x = self.dropout2(x)
# add residual
if self.conv1.in_channels != self.conv2.out_channels:
residual = self.conv3(residual)
x = x + residual
return x
class _TCNModule(PLPastCovariatesModule):
def __init__(
self,
input_size: int,
kernel_size: int,
num_filters: int,
num_layers: Optional[int],
dilation_base: int,
weight_norm: bool,
target_size: int,
nr_params: int,
target_length: int,
dropout: float,
**kwargs
):
"""PyTorch module implementing a dilated TCN module used in `TCNModel`.
Parameters
----------
input_size
The dimensionality of the input time series.
target_size
The dimensionality of the output time series.
nr_params
The number of parameters of the likelihood (or 1 if no likelihood is used).
target_length
Number of time steps the torch module will predict into the future at once.
kernel_size
The size of every kernel in a convolutional layer.
num_filters
The number of filters in a convolutional layer of the TCN.
num_layers
The number of convolutional layers.
weight_norm
Boolean value indicating whether to use weight normalization.
dilation_base
The base of the exponent that will determine the dilation on every level.
dropout
The dropout rate for every convolutional layer.
**kwargs
all parameters required for :class:`darts.models.forecasting.pl_forecasting_module.PLForecastingModule`
base class.
Inputs
------
x of shape `(batch_size, input_chunk_length, input_size)`
Tensor containing the features of the input sequence.
Outputs
-------
y of shape `(batch_size, input_chunk_length, target_size, nr_params)`
Tensor containing the predictions of the next 'output_chunk_length' points in the last
'output_chunk_length' entries of the tensor. The entries before contain the data points
leading up to the first prediction, all in chronological order.
"""
super().__init__(**kwargs)
# Defining parameters
self.input_size = input_size
self.n_filters = num_filters
self.kernel_size = kernel_size
self.target_length = target_length
self.target_size = target_size
self.nr_params = nr_params
self.dilation_base = dilation_base
# If num_layers is not passed, compute number of layers needed for full history coverage
if num_layers is None and dilation_base > 1:
num_layers = math.ceil(
math.log(
(self.input_chunk_length - 1)
* (dilation_base - 1)
/ (kernel_size - 1)
/ 2
+ 1,
dilation_base,
)
)
logger.info("Number of layers chosen: " + str(num_layers))
elif num_layers is None:
num_layers = math.ceil(
(self.input_chunk_length - 1) / (kernel_size - 1) / 2
)
logger.info("Number of layers chosen: " + str(num_layers))
self.num_layers = num_layers
# Building TCN module
self.res_blocks_list = []
for i in range(num_layers):
res_block = _ResidualBlock(
num_filters=num_filters,
kernel_size=kernel_size,
dilation_base=dilation_base,
dropout=dropout,
weight_norm=weight_norm,
nr_blocks_below=i,
num_layers=num_layers,
input_size=self.input_size,
target_size=target_size * nr_params,
)
self.res_blocks_list.append(res_block)
self.res_blocks = nn.ModuleList(self.res_blocks_list)
@io_processor
def forward(self, x_in: Tuple):
x, _ = x_in
# data is of size (batch_size, input_chunk_length, input_size)
batch_size = x.size(0)
x = x.transpose(1, 2)
for res_block in self.res_blocks_list:
x = res_block(x)
x = x.transpose(1, 2)
x = x.view(
batch_size, self.input_chunk_length, self.target_size, self.nr_params
)
return x
@property
def first_prediction_index(self) -> int:
return -self.output_chunk_length
[docs]class TCNModel(PastCovariatesTorchModel):
def __init__(
self,
input_chunk_length: int,
output_chunk_length: int,
output_chunk_shift: int = 0,
kernel_size: int = 3,
num_filters: int = 3,
num_layers: Optional[int] = None,
dilation_base: int = 2,
weight_norm: bool = False,
dropout: float = 0.2,
**kwargs
):
"""Temporal Convolutional Network Model (TCN).
This is an implementation of a dilated TCN used for forecasting, inspired from [1]_.
This model supports past covariates (known for `input_chunk_length` points before prediction time).
Parameters
----------
input_chunk_length
Number of time steps in the past to take as a model input (per chunk). Applies to the target
series, and past and/or future covariates (if the model supports it).
output_chunk_length
Number of time steps predicted at once (per chunk) by the internal model. Also, the number of future values
from future covariates to use as a model input (if the model supports future covariates). It is not the same
as forecast horizon `n` used in `predict()`, which is the desired number of prediction points generated
using either a one-shot- or autoregressive forecast. Setting `n <= output_chunk_length` prevents
auto-regression. This is useful when the covariates don't extend far enough into the future, or to prohibit
the model from using future values of past and / or future covariates for prediction (depending on the
model's covariate support).
output_chunk_shift
Optionally, the number of steps to shift the start of the output chunk into the future (relative to the
input chunk end). This will create a gap between the input and output. If the model supports
`future_covariates`, the future values are extracted from the shifted output chunk. Predictions will start
`output_chunk_shift` steps after the end of the target `series`. If `output_chunk_shift` is set, the model
cannot generate autoregressive predictions (`n > output_chunk_length`).
kernel_size
The size of every kernel in a convolutional layer.
num_filters
The number of filters in a convolutional layer of the TCN.
weight_norm
Boolean value indicating whether to use weight normalization.
dilation_base
The base of the exponent that will determine the dilation on every level.
num_layers
The number of convolutional layers.
dropout
The dropout rate for every convolutional layer. This is compatible with Monte Carlo dropout
at inference time for model uncertainty estimation (enabled with ``mc_dropout=True`` at
prediction time).
**kwargs
Optional arguments to initialize the pytorch_lightning.Module, pytorch_lightning.Trainer, and
Darts' :class:`TorchForecastingModel`.
loss_fn
PyTorch loss function used for training.
This parameter will be ignored for probabilistic models if the ``likelihood`` parameter is specified.
Default: ``torch.nn.MSELoss()``.
likelihood
One of Darts' :meth:`Likelihood <darts.utils.likelihood_models.Likelihood>` models to be used for
probabilistic forecasts. Default: ``None``.
torch_metrics
A torch metric or a ``MetricCollection`` used for evaluation. A full list of available metrics can be found
at https://torchmetrics.readthedocs.io/en/latest/. Default: ``None``.
optimizer_cls
The PyTorch optimizer class to be used. Default: ``torch.optim.Adam``.
optimizer_kwargs
Optionally, some keyword arguments for the PyTorch optimizer (e.g., ``{'lr': 1e-3}``
for specifying a learning rate). Otherwise the default values of the selected ``optimizer_cls``
will be used. Default: ``None``.
lr_scheduler_cls
Optionally, the PyTorch learning rate scheduler class to be used. Specifying ``None`` corresponds
to using a constant learning rate. Default: ``None``.
lr_scheduler_kwargs
Optionally, some keyword arguments for the PyTorch learning rate scheduler. Default: ``None``.
use_reversible_instance_norm
Whether to use reversible instance normalization `RINorm` against distribution shift as shown in [2]_.
It is only applied to the features of the target series and not the covariates.
batch_size
Number of time series (input and output sequences) used in each training pass. Default: ``32``.
n_epochs
Number of epochs over which to train the model. Default: ``100``.
model_name
Name of the model. Used for creating checkpoints and saving tensorboard data. If not specified,
defaults to the following string ``"YYYY-mm-dd_HH_MM_SS_torch_model_run_PID"``, where the initial part
of the name is formatted with the local date and time, while PID is the processed ID (preventing models
spawned at the same time by different processes to share the same model_name). E.g.,
``"2021-06-14_09_53_32_torch_model_run_44607"``.
work_dir
Path of the working directory, where to save checkpoints and Tensorboard summaries.
Default: current working directory.
log_tensorboard
If set, use Tensorboard to log the different parameters. The logs will be located in:
``"{work_dir}/darts_logs/{model_name}/logs/"``. Default: ``False``.
nr_epochs_val_period
Number of epochs to wait before evaluating the validation loss (if a validation
``TimeSeries`` is passed to the :func:`fit()` method). Default: ``1``.
force_reset
If set to ``True``, any previously-existing model with the same name will be reset (all checkpoints will
be discarded). Default: ``False``.
save_checkpoints
Whether to automatically save the untrained model and checkpoints from training.
To load the model from checkpoint, call :func:`MyModelClass.load_from_checkpoint()`, where
:class:`MyModelClass` is the :class:`TorchForecastingModel` class that was used (such as :class:`TFTModel`,
:class:`NBEATSModel`, etc.). If set to ``False``, the model can still be manually saved using
:func:`save()` and loaded using :func:`load()`. Default: ``False``.
add_encoders
A large number of past and future covariates can be automatically generated with `add_encoders`.
This can be done by adding multiple pre-defined index encoders and/or custom user-made functions that
will be used as index encoders. Additionally, a transformer such as Darts' :class:`Scaler` can be added to
transform the generated covariates. This happens all under one hood and only needs to be specified at
model creation.
Read :meth:`SequentialEncoder <darts.dataprocessing.encoders.SequentialEncoder>` to find out more about
``add_encoders``. Default: ``None``. An example showing some of ``add_encoders`` features:
.. highlight:: python
.. code-block:: python
def encode_year(idx):
return (idx.year - 1950) / 50
add_encoders={
'cyclic': {'future': ['month']},
'datetime_attribute': {'future': ['hour', 'dayofweek']},
'position': {'past': ['relative'], 'future': ['relative']},
'custom': {'past': [encode_year]},
'transformer': Scaler(),
'tz': 'CET'
}
..
random_state
Control the randomness of the weights initialization. Check this
`link <https://scikit-learn.org/stable/glossary.html#term-random_state>`_ for more details.
Default: ``None``.
pl_trainer_kwargs
By default :class:`TorchForecastingModel` creates a PyTorch Lightning Trainer with several useful presets
that performs the training, validation and prediction processes. These presets include automatic
checkpointing, tensorboard logging, setting the torch device and more.
With ``pl_trainer_kwargs`` you can add additional kwargs to instantiate the PyTorch Lightning trainer
object. Check the `PL Trainer documentation
<https://pytorch-lightning.readthedocs.io/en/stable/common/trainer.html>`_ for more information about the
supported kwargs. Default: ``None``.
Running on GPU(s) is also possible using ``pl_trainer_kwargs`` by specifying keys ``"accelerator",
"devices", and "auto_select_gpus"``. Some examples for setting the devices inside the ``pl_trainer_kwargs``
dict:rgs``
dict:
- ``{"accelerator": "cpu"}`` for CPU,
- ``{"accelerator": "gpu", "devices": [i]}`` to use only GPU ``i`` (``i`` must be an integer),
- ``{"accelerator": "gpu", "devices": -1, "auto_select_gpus": True}`` to use all available GPUS.
For more info, see here:
https://pytorch-lightning.readthedocs.io/en/stable/common/trainer.html#trainer-flags , and
https://pytorch-lightning.readthedocs.io/en/stable/accelerators/gpu_basic.html#train-on-multiple-gpus
With parameter ``"callbacks"`` you can add custom or PyTorch-Lightning built-in callbacks to Darts'
:class:`TorchForecastingModel`. Below is an example for adding EarlyStopping to the training process.
The model will stop training early if the validation loss `val_loss` does not improve beyond
specifications. For more information on callbacks, visit:
`PyTorch Lightning Callbacks
<https://pytorch-lightning.readthedocs.io/en/stable/extensions/callbacks.html>`_
.. highlight:: python
.. code-block:: python
from pytorch_lightning.callbacks.early_stopping import EarlyStopping
# stop training when validation loss does not decrease more than 0.05 (`min_delta`) over
# a period of 5 epochs (`patience`)
my_stopper = EarlyStopping(
monitor="val_loss",
patience=5,
min_delta=0.05,
mode='min',
)
pl_trainer_kwargs={"callbacks": [my_stopper]}
..
Note that you can also use a custom PyTorch Lightning Trainer for training and prediction with optional
parameter ``trainer`` in :func:`fit()` and :func:`predict()`.
show_warnings
whether to show warnings raised from PyTorch Lightning. Useful to detect potential issues of
your forecasting use case. Default: ``False``.
References
----------
.. [1] https://arxiv.org/abs/1803.01271
.. [2] T. Kim et al. "Reversible Instance Normalization for Accurate Time-Series Forecasting against
Distribution Shift", https://openreview.net/forum?id=cGDAkQo1C0p
Examples
--------
>>> from darts.datasets import WeatherDataset
>>> from darts.models import TCNModel
>>> series = WeatherDataset().load()
>>> # predicting atmospheric pressure
>>> target = series['p (mbar)'][:100]
>>> # optionally, use past observed rainfall (pretending to be unknown beyond index 100)
>>> past_cov = series['rain (mm)'][:100]
>>> # `output_chunk_length` must be strictly smaller than `input_chunk_length`
>>> model = TCNModel(
>>> input_chunk_length=12,
>>> output_chunk_length=6,
>>> n_epochs=20,
>>> )
>>> model.fit(target, past_covariates=past_cov)
>>> pred = model.predict(6)
>>> pred.values()
array([[-80.48476824],
[-80.47896667],
[-41.77135603],
[-41.76158729],
[-41.76854107],
[-41.78166819]])
.. note::
`DeepTCN example notebook <https://unit8co.github.io/darts/examples/09-DeepTCN-examples.html>`_ presents
techniques that can be used to improve the forecasts quality compared to this simple usage example.
"""
raise_if_not(
kernel_size < input_chunk_length,
"The kernel size must be strictly smaller than the input length.",
logger,
)
raise_if_not(
output_chunk_length < input_chunk_length,
"The output length must be strictly smaller than the input length",
logger,
)
super().__init__(**self._extract_torch_model_params(**self.model_params))
# extract pytorch lightning module kwargs
self.pl_module_params = self._extract_pl_module_params(**self.model_params)
self.kernel_size = kernel_size
self.num_filters = num_filters
self.num_layers = num_layers
self.dilation_base = dilation_base
self.dropout = dropout
self.weight_norm = weight_norm
@property
def supports_multivariate(self) -> bool:
return True
def _create_model(self, train_sample: Tuple[torch.Tensor]) -> torch.nn.Module:
# samples are made of (past_target, past_covariates, future_target)
input_dim = train_sample[0].shape[1] + (
train_sample[1].shape[1] if train_sample[1] is not None else 0
)
output_dim = train_sample[-1].shape[1]
nr_params = 1 if self.likelihood is None else self.likelihood.num_parameters
return _TCNModule(
input_size=input_dim,
target_size=output_dim,
nr_params=nr_params,
kernel_size=self.kernel_size,
num_filters=self.num_filters,
num_layers=self.num_layers,
dilation_base=self.dilation_base,
target_length=self.output_chunk_length,
dropout=self.dropout,
weight_norm=self.weight_norm,
**self.pl_module_params,
)
def _build_train_dataset(
self,
target: Sequence[TimeSeries],
past_covariates: Optional[Sequence[TimeSeries]],
future_covariates: Optional[Sequence[TimeSeries]],
max_samples_per_ts: Optional[int],
) -> PastCovariatesShiftedDataset:
return PastCovariatesShiftedDataset(
target_series=target,
covariates=past_covariates,
length=self.input_chunk_length,
shift=self.output_chunk_length + self.output_chunk_shift,
max_samples_per_ts=max_samples_per_ts,
use_static_covariates=self.uses_static_covariates,
)