remove shifts

etna/models/nn/patchts.py

@@ -112,7 +112,7 @@ def __init__(
  PositionalEncoding(d_model=self.hidden_size),
  nn.TransformerEncoder(encoder_layers, self.num_layers)
  )
- self.max_patch_num = (encoder_length - 1 - self.patch_len) // self.stride + 1
+ self.max_patch_num = (encoder_length - self.patch_len) // self.stride + 1
  self.projection = nn.Sequential(nn.Flatten(start_dim=-2),
  nn.Linear(in_features=self.hidden_size * self.max_patch_num, out_features=1))
 
@@ -132,15 +132,13 @@ def forward(self, x: PatchTSBatch, *args, **kwargs): # type: ignore
  :
  forecast with shape (batch_size, decoder_length, 1)
  """
- print(x)
  encoder_real = x["encoder_real"].float() # (batch_size, encoder_length-1, input_size)
  decoder_real = x["decoder_real"].float() # (batch_size, decoder_length, input_size)
  decoder_length = decoder_real.shape[1]
  outputs = []
  x = encoder_real
  for i in range(decoder_length):
  pred = self._get_prediction(x)
- print(pred)
  outputs.append(pred)
  x = torch.cat((x[:, 1:, :], torch.unsqueeze(pred, dim=1)), dim=1)
 
@@ -150,7 +148,7 @@ def forward(self, x: PatchTSBatch, *args, **kwargs): # type: ignore
  return forecast
 
  def _get_prediction(self, x: torch.Tensor) -> torch.Tensor:
- x = x.permute(0, 2, 1) # (batch_size, input_size, encoder_length + decoder_length - 1)
+ x = x.permute(0, 2, 1) # (batch_size, input_size, encoder_length)
  # do patching
  x = x.unfold(dimension=-1, size=self.patch_len,
  step=self.stride) # (batch_size, input_size, patch_num, patch_len)
@@ -173,7 +171,7 @@ def step(self, batch: PatchTSBatch, *args, **kwargs): # type: ignore
  :
  loss, true_target, prediction_target
  """
- encoder_real = batch["encoder_real"].float() # (batch_size, encoder_length-1, input_size)
+ encoder_real = batch["encoder_real"].float() # (batch_size, encoder_length, input_size)
  decoder_real = batch["decoder_real"].float() # (batch_size, decoder_length, input_size)
 
  decoder_target = batch["decoder_target"].float() # (batch_size, decoder_length, 1)
@@ -196,11 +194,7 @@ def step(self, batch: PatchTSBatch, *args, **kwargs): # type: ignore
  def make_samples(self, df: pd.DataFrame, encoder_length: int, decoder_length: int) -> Iterator[dict]:
  """Make samples from segment DataFrame."""
  values_real = (
- df.select_dtypes(include=[np.number])
- .assign(target_shifted=df["target"].shift(1))
- .drop(["target"], axis=1)
- .pipe(lambda x: x[["target_shifted"] + [i for i in x.columns if i != "target_shifted"]])
- .values
+ df.select_dtypes(include=[np.number]).values
  )
  values_target = df["target"].values
  segment = df["segment"].values[0]
@@ -232,10 +226,10 @@ def _make(
 
  # Get shifted target and concatenate it with real values features
  sample["encoder_real"] = values_real[start_idx: start_idx + encoder_length]
- sample["encoder_real"] = sample["encoder_real"][1:]
+ sample["encoder_real"] = sample["encoder_real"]
 
  target = values_target[start_idx: start_idx + encoder_length + decoder_length].reshape(-1, 1)
- sample["encoder_target"] = target[1:encoder_length]
+ sample["encoder_target"] = target[:encoder_length]
  sample["decoder_target"] = target[encoder_length:]
 
  sample["segment"] = segment
@@ -278,8 +272,8 @@ def __init__(
  nhead: int = 16,
  lr: float = 1e-3,
  loss: "torch.nn.Module" = nn.MSELoss(),
- train_batch_size: int = 64,
- test_batch_size: int = 64,
+ train_batch_size: int = 128,
+ test_batch_size: int = 128,
  optimizer_params: Optional[dict] = None,
  trainer_params: Optional[dict] = None,
  train_dataloader_params: Optional[dict] = None,

tests/test_models/nn/test_patchts.py

@@ -11,8 +11,8 @@
  "horizon",
  [
  8,
- # 13,
- # 15
+ 13,
+ 15
  ],
 )
 def test_patchts_model_run_weekly_overfit_with_scaler_small_patch(ts_dataset_weekly_function_with_horizon, horizon):
@@ -30,6 +30,7 @@ def test_patchts_model_run_weekly_overfit_with_scaler_small_patch(ts_dataset_wee
  future.inverse_transform([std])
 
  mae = MAE("macro")
+ print(mae(ts_test, future))
  assert mae(ts_test, future) < 0.9