MODULE REFERENCE

core

anomx.data.views.core

Standardized materialized views over an ObservationSet.

The source observation set stays canonical; algorithms request the materialization that matches their capabilities:

` ObservationSet ├── TabularView n_samples × n_features matrix ├── SequenceView per-sequence ordered frames ├── WindowView fixed-size sliding windows ├── TimeSeriesView time-indexed frame (darts-compatible) └── GraphView node features plus relational overlay `

classView source ↗

BaseObservationView

BaseObservationView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>)

Common surface for all materialized observation views.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`

Members

sourceattribute

# BaseObservationView.source Source ↗

source: ObservationSet

metadataattribute

# BaseObservationView.metadata Source ↗

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

classView source ↗

GraphView

GraphView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>, nodes: pd.DataFrame = <factory: pd.DataFrame>, relations: RelationSet | None = None)

Bases: BaseObservationView

Node feature matrix combined with the relational overlay.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`nodes`	`pd.DataFrame`	`<factory: pd.DataFrame>`
`relations`	`RelationSet \| None`	`None`

Members

nodesattribute

# GraphView.nodes Source ↗

nodes: pd.DataFrame = field(default_factory=pd.DataFrame)

relationsattribute

# GraphView.relations Source ↗

relations: RelationSet | None = None

node_countproperty

# GraphView.node_count Source ↗

node_count: int

Returns int

classView source ↗

ObservationWindow

ObservationWindow(window_id: int, group: Any, start_position: int, end_position: int, frame: pd.DataFrame)

One fixed-size window sliced out of an ordered observation frame.

Signature parameters
Parameter	Type	Default
`window_id`	`int`	`required`
`group`	`Any`	`required`
`start_position`	`int`	`required`
`end_position`	`int`	`required`
`frame`	`pd.DataFrame`	`required`

Members

window_idattribute

# ObservationWindow.window_id Source ↗

window_id: int

groupattribute

# ObservationWindow.group Source ↗

group: Any

start_positionattribute

# ObservationWindow.start_position Source ↗

start_position: int

end_positionattribute

# ObservationWindow.end_position Source ↗

end_position: int

frameattribute

# ObservationWindow.frame Source ↗

frame: pd.DataFrame

classView source ↗

SequenceView

SequenceView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>, sequences: dict[Any, pd.DataFrame] = <factory: dict>)

Bases: BaseObservationView

Ordered frames grouped by sequence (or entity) identifier.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`sequences`	`dict[Any, pd.DataFrame]`	`<factory: dict>`

Members

sequencesattribute

# SequenceView.sequences Source ↗

sequences: dict[Any, pd.DataFrame] = field(default_factory=dict)

sequence_countproperty

# SequenceView.sequence_count Source ↗

sequence_count: int

Returns int

lengthsmethod

# SequenceView.lengths Source ↗

lengths() -> dict[Any, int]

Returns dict[Any, int]

classView source ↗

TabularView

TabularView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>, matrix: pd.DataFrame = <factory: pd.DataFrame>)

Bases: BaseObservationView

Unordered n_samples × n_features matrix materialization.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`matrix`	`pd.DataFrame`	`<factory: pd.DataFrame>`

Members

matrixattribute

# TabularView.matrix Source ↗

matrix: pd.DataFrame = field(default_factory=pd.DataFrame)

to_numpymethod

# TabularView.to_numpy Source ↗

to_numpy() -> Any

Returns Any

classView source ↗

TimeSeriesView

TimeSeriesView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>, frame: pd.DataFrame = <factory: pd.DataFrame>, time_column: str = 'timestamp', value_columns: list[str] = <factory: list>)

Bases: BaseObservationView

Time-indexed materialization for temporal models.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`frame`	`pd.DataFrame`	`<factory: pd.DataFrame>`
`time_column`	`str`	`'timestamp'`
`value_columns`	`list[str]`	`<factory: list>`

Members

frameattribute

# TimeSeriesView.frame Source ↗

frame: pd.DataFrame = field(default_factory=pd.DataFrame)

time_columnattribute

# TimeSeriesView.time_column Source ↗

time_column: str = 'timestamp'

value_columnsattribute

# TimeSeriesView.value_columns Source ↗

value_columns: list[str] = field(default_factory=list)

to_dartsmethod

# TimeSeriesView.to_darts Source ↗

to_darts() -> Any

Convert into a darts.TimeSeries (requires the darts extra).

Returns Any

classView source ↗

WindowView

WindowView(source: ObservationSet, metadata: dict[str, Any] = <factory: dict>, windows: list[ObservationWindow] = <factory: list>, window_size: int = 0, stride: int = 1)

Bases: BaseObservationView

Sliding fixed-size windows over ordered observations.

Signature parameters
Parameter	Type	Default
`source`	`ObservationSet`	`required`
`metadata`	`dict[str, Any]`	`<factory: dict>`
`windows`	`list[ObservationWindow]`	`<factory: list>`
`window_size`	`int`	`0`
`stride`	`int`	`1`

Members

windowsattribute

# WindowView.windows Source ↗

windows: list[ObservationWindow] = field(default_factory=list)

window_sizeattribute

# WindowView.window_size Source ↗

window_size: int = 0

strideattribute

# WindowView.stride Source ↗

stride: int = 1

window_countproperty

# WindowView.window_count Source ↗

window_count: int

Returns int