MODULE REFERENCE

datasets

anomx.data.datasets

Core dataset abstractions.

ChannelMetadata

ChannelMetadata(name: str, unit: str | None = None, description: str | None = None, tags: tuple[str, ...] = ())

Metadata describing one logical time-series channel.

Signature parameters
Parameter	Type	Default
`name`	`str`	`required`
`unit`	`str \| None`	`None`
`description`	`str \| None`	`None`
`tags`	`tuple[str, ...]`	`()`

Members

nameattribute

# ChannelMetadata.name Source ↗

name: str

unitattribute

# ChannelMetadata.unit Source ↗

unit: str | None = None

descriptionattribute

# ChannelMetadata.description Source ↗

description: str | None = None

tagsattribute

# ChannelMetadata.tags Source ↗

tags: tuple[str, ...] = ()

classView source ↗

TimeSeriesDataset

TimeSeriesDataset(data: pd.DataFrame, target_columns: tuple[str, ...], name: str | None = None, metadata: dict[str, Any] = <factory: dict>, channels: dict[str, ChannelMetadata] = <factory: dict>)

A typed container for one or more aligned time-series channels.

The container keeps the public library independent from the Anomx platform's storage layer while preserving the metadata needed for auditable pipelines.

Signature parameters
Parameter	Type	Default
`data`	`pd.DataFrame`	`required`
`target_columns`	`tuple[str, ...]`	`required`
`name`	`str \| None`	`None`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`channels`	`dict[str, ChannelMetadata]`	`<factory: dict>`

Members

dataattribute

# TimeSeriesDataset.data Source ↗

data: pd.DataFrame

target_columnsattribute

# TimeSeriesDataset.target_columns Source ↗

target_columns: tuple[str, ...]

nameattribute

# TimeSeriesDataset.name Source ↗

name: str | None = None

metadataattribute

# TimeSeriesDataset.metadata Source ↗

metadata: dict[str, Any] = field(default_factory=dict)

channelsattribute

# TimeSeriesDataset.channels Source ↗

channels: dict[str, ChannelMetadata] = field(default_factory=dict)

from_dataframeclassmethod

# TimeSeriesDataset.from_dataframe Source ↗

from_dataframe(frame: pd.DataFrame, *, target_columns: str | list[str] | tuple[str, ...], time_column: str | None = None, name: str | None = None, metadata: dict[str, Any] | None = None) -> TimeSeriesDataset

Create a dataset from a DataFrame.

Args

frame: Source frame.
target_columns: Column or columns modeled by detectors and forecasters.
time_column: Optional column to convert into the DatetimeIndex.
name: Optional human-readable dataset name.
metadata: Optional metadata carried through pipelines.

Signature parameters
Parameter	Type	Default
`frame`	`pd.DataFrame`	`required`
`target_columns`keyword only	`str \| list[str] \| tuple[str, ...]`	`required`
`time_column`keyword only	`str \| None`	`None`
`name`keyword only	`str \| None`	`None`
`metadata`keyword only	`dict[str, Any] \| None`	`None`

Returns TimeSeriesDataset

frequencyproperty

# TimeSeriesDataset.frequency Source ↗

frequency: str | None

Infer the dataset frequency when pandas can determine it.

Returns str | None

targetproperty

# TimeSeriesDataset.target Source ↗

target: pd.Series | pd.DataFrame

Return the modeled target column or columns.

Returns pd.Series | pd.DataFrame

to_dataframemethod

# TimeSeriesDataset.to_dataframe Source ↗

to_dataframe() -> pd.DataFrame

Return a copy of the underlying DataFrame.

Returns pd.DataFrame

train_test_splitmethod

# TimeSeriesDataset.train_test_split Source ↗

train_test_split(test_size: float | int) -> tuple[TimeSeriesDataset, TimeSeriesDataset]

Split the dataset into chronological train and test portions.

Signature parameters
Parameter	Type	Default
`test_size`	`float \| int`	`required`

Returns tuple[TimeSeriesDataset, TimeSeriesDataset]

resamplemethod

# TimeSeriesDataset.resample Source ↗

resample(rule: str, *, method: str = 'mean') -> TimeSeriesDataset

Return a resampled dataset using a pandas aggregation method.

Signature parameters
Parameter	Type	Default
`rule`	`str`	`required`
`method`keyword only	`str`	`'mean'`

Returns TimeSeriesDataset