High-level graph layer corresponding to a single stage of a multi-stage inter-partition shuffle operation.

Stage: (shuffle-group) -> (shuffle-split) -> (shuffle-join)

Parameters

name : str

Name of new (partially) shuffled collection.

column : str or list of str

Column(s) to be used to map rows to output partitions (by hashing).

inputs : list of tuples

Each tuple dictates the data movement for a specific partition.

stage : int

Index of the current shuffle stage.

npartitions : int

Number of output partitions for the full (multi-stage) shuffle.

npartitions_input : int

Number of partitions in the original (un-shuffled) DataFrame.

k : int

A partition is split into this many groups during each stage.

ignore_index: bool, default False :

Ignore index during shuffle. If True , performance may improve, but index values will not be preserved.

name_input : str

Name of input collection.

meta_input : pd.DataFrame-like object

Empty metadata of input collection.

parts_out : list of int (optional)

List of required output-partition indices.

annotations : dict (optional)

Layer annotations

Shuffle-stage HighLevelGraph layer

Examples

See :

Local connectivity graph

Hover to see nodes names; edges to Self not shown, Caped at 50 nodes.

Using a canvas is more power efficient and can get hundred of nodes ; but does not allow hyperlinks; , arrows or text (beyond on hover)

SVG is more flexible but power hungry; and does not scale well to 50 + nodes.

All aboves nodes referred to, (or are referred from) current nodes; Edges from Self to other have been omitted (or all nodes would be connected to the central node "self" which is not useful). Nodes are colored by the library they belong to, and scaled with the number of references pointing them

---

File: /dask/layers.py#633
type: <class 'abc.ABCMeta'>
Commit: