timesead.data.statistics

Functions

`get_data_all`(→ torch.Tensor)	Extract data matrix from dataset.
`compute_whiten_matrix`(→ torch.Tensor)	Compute whiten matrix of dataset.
`compute_feature_statistics`(...)	Compute statistics of each feature.
`compute_feature_mean_std`(...)	Compute the mean and standard deviation of each feature.
`compute_ts_statisitcs`(→ Tuple[Optional[torch.Tensor], ...)
`training_means`(dataset)
`compute_anomaly_positions`(→ List[int])	Computes the positions of anomalies in the dataset.
`compute_anomaly_lengths`(→ List[int])	Computes the length of each anomalous window in the dataset.
`compute_total_time_steps`(→ int)	Compute the total amount of time steps in the dataset (normal + anormal)

Module Contents

timesead.data.statistics.get_data_all(dataset: torch.utils.data.Dataset, take_fd: bool = False) → torch.Tensor

Extract data matrix from dataset.

Parameters:

dataset (torch.utils.data.Dataset take_id: bool) – Time-series dataset. take_fd: flag, to take first differences.
take_fd (bool)

Returns:

data matrix

Return type:

torch.Tensor

timesead.data.statistics.compute_whiten_matrix(dataset: torch.utils.data.Dataset, take_fd: bool = False) → torch.Tensor

Compute whiten matrix of dataset.

Parameters:

dataset (torch.utils.data.Dataset) – Time-series dataset.
take_fd (bool)

Returns:

whiten matrix

Return type:

torch.Tensor

timesead.data.statistics.compute_feature_statistics(dataset: torch.utils.data.Dataset, take_fd: bool = False) → Tuple[torch.Tensor | None, torch.Tensor | None, torch.Tensor | None, torch.Tensor | None, torch.Tensor, torch.Tensor]

Compute statistics of each feature.

Parameters:

dataset (torch.utils.data.Dataset) – Time-series dataset.
take_fd (bool)

Returns:

Normal mean, normal std, anomaly mean, anomaly std, minimum, and maximum

Return type:

Tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], torch.Tensor, torch.Tensor]

timesead.data.statistics.compute_feature_mean_std(dataset: timesead.data.dataset.BaseTSDataset, take_fd: bool = False) → Tuple[torch.Tensor | None, torch.Tensor | None, torch.Tensor | None, torch.Tensor | None]

Compute the mean and standard deviation of each feature. :param dataset: Time-series dataset. :type dataset: torch.utils.data.Dataset :return: Normal mean, normal std, anomaly mean, anomaly std :rtype: Tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor]]

Parameters:

dataset (timesead.data.dataset.BaseTSDataset)
take_fd (bool)

Return type:

Tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor]]

timesead.data.statistics.compute_ts_statisitcs(timeseries: torch.Tensor, targets: torch.Tensor, dim_order: str = 'tf') → Tuple[torch.Tensor | None, torch.Tensor | None, torch.Tensor | None, torch.Tensor | None, torch.Tensor, torch.Tensor]

Parameters:

timeseries (torch.Tensor)
targets (torch.Tensor)
dim_order (str)

Return type:

Tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor], torch.Tensor, torch.Tensor]

timesead.data.statistics.training_means(dataset)

timesead.data.statistics.compute_anomaly_positions(dataset: torch.utils.data.Dataset) → List[int]

Computes the positions of anomalies in the dataset.

Parameters:: dataset (torch.utils.data.Dataset) – Dataset to compute the statistics of.
Returns:: List of all the relative positions of anomalies in the dataset.
Return type:: List[int]

timesead.data.statistics.compute_anomaly_lengths(dataset: torch.utils.data.Dataset) → List[int]

Computes the length of each anomalous window in the dataset.

Parameters:: dataset (torch.utils.data.Dataset) – Dataset to compute the statistics of.
Returns:: List of lengths of anomalies in the dataset.
Return type:: List[int]

timesead.data.statistics.compute_total_time_steps(dataset: torch.utils.data.Dataset) → int

Compute the total amount of time steps in the dataset (normal + anormal)

Parameters:: dataset (torch.utils.data.Dataset) – Dataset to compute the statistics of.
Returns:: Number of time steps in the dataset
Return type:: int