tlc.core.objects.tables.from_table.umap_table

A Table where a column from the input Table has been dimensionally reduced by the UMAP algorithm.

Module Contents

Classes

Class

Description

UMAPTable

A procedural table where a column in the input table column has been has dimensionally reduced by the UMAP algorithm.

Data

Data

Description

umap

API

umap = _lazy_import(...)
class UMAPTable(
*,
url: Url | None = None,
created: str | None = None,
description: str | None = None,
row_cache_url: Url | None = None,
row_cache_populated: bool | None = None,
input_table_url: Url | Table | None = None,
source_embedding_column: str | None = None,
target_embedding_column: str | None = None,
retain_source_embedding_column: bool | None = None,
standard_scaler_normalize: bool | None = None,
n_components: int | None = None,
n_neighbors: int | None = None,
metric: str | None = None,
min_dist: float | None = None,
n_jobs: int | None = None,
fit_table_url: Table | Url | None = None,
model_url: Url | None = None,
init_parameters: Any = None,
random_state: int | None = None,
input_tables: list[Url] | None = None,
)

Bases: tlc.core.objects.tables.from_table.dimensional_reduction_table._DimensionalReductionTable

A procedural table where a column in the input table column has been has dimensionally reduced by the UMAP algorithm.

Creates a derived table with an (additional) UMAP-ed column based on input column and wanted dimensionality.

Parameters:

  • input_table_url – The input table to apply UMAP to

  • source_embedding_column – The column in the input table to apply UMAP to

  • target_embedding_column – The name of the new column to create in the output table

  • retain_source_embedding_column – Whether to retain the source column in the UMAP table, defaults to False

  • standard_scaler_normalize – Whether to apply the sklearn standard scaler to input before mapping, defaults to False

  • n_components – The dimension of the output embedding

  • n_neighbors – The number of neighbors to use to approximate the manifold structure

  • metric – The metric to use to compute distances in high dimensional space

  • min_dist – The minimum distance between points in the low dimensional embedding

  • n_jobs – The number of threads to use for the reduction. If set to anything other than 1, the random_state parameter of the UMAP algorithm is set to None, which means that the results will not be deterministic.

  • fit_table_url – The table to use for fitting the UMAP transform, if not specified the input table is used

  • model_url – The URL to store/load the UMAP model file. If empty, no model is saved.

  • random_state – The random state to use for the reduction

algorithm_name = UMAP
property seed: list[int]