anchor_python_visualization.projection

Methods for projecting a feature space to lower dimensionality.

Package Contents

Classes

Projector

Projects the feature-space to lower dimensionality.

Functions

create_projector(...)

Creates a projection method from an identifier.

Attributes

DEFAULT_IDENTIFIER	The default choice to use in IDENTIFIERS.
IDENTIFIERS	Unique strings to use as command-line-arguments to select a Projector.

anchor_python_visualization.projection.DEFAULT_IDENTIFIER = 't-sne'

The default choice to use in IDENTIFIERS.

anchor_python_visualization.projection.IDENTIFIERS = ['t-sne', 'pca', 'none']

Unique strings to use as command-line-arguments to select a Projector.

All are lower-case.

anchor_python_visualization.projection.create_projector(identifier: str) → anchor_python_visualization.projection.projector.Projector | None

Creates a projection method from an identifier.

Parameters:

identifier – string that is one of IDENTIFIERS, case-insensitive.

Returns:

a newly created projection method, or none at all.

class anchor_python_visualization.projection.Projector

Bases: abc.ABC

Projects the feature-space to lower dimensionality.

project(features: pandas.DataFrame) → pandas.DataFrame

Performs projection, while preserving a data-frame with identical row names.

Parameters:

• features – data_frame containing only numerical embeddings (as columns) and with labelled

• row.names –

Returns:

a data-frame of embeddings with identical order and row names, but changed columns.