Dimensionality Reduction API

tslpp

Module for analysis with LPP - Used by TS-LPP as well

AtomicAI.dim_reduction.tslpp.compute_lpp(descriptor_data, graph_nearest_neighbors, sigma, n_components_target)[source]

Locally preserving projection.

Parameters:

  • n_components_target (int)

  • sigma (float)

  • descriptor_data (float64) – 2D array (typically, numpy.array)

  • graph_nearest_neighbors (int) – number (knn)

Returns:

AtomicAI.dim_reduction.tslpp.new_centering(reference_data, my_data)[source]

Keep only columns that have some variance.

Parameters:

  • reference_data

  • my_data

Returns:

AtomicAI.dim_reduction.tslpp.perform_lpp(descriptor_data, graph_nearest_neighbors=None, sigma=None, final_reduced_dimensions=None, intermediate_dimensions=None)[source]
AtomicAI.dim_reduction.tslpp.tslpp(features, reduced_dim, sigma, intermediate_dimensions)[source]

lpp

Module for analysis with LPP - Used by TS-LPP as well

AtomicAI.dim_reduction.lpp.compute_lpp(descriptor_data, graph_nearest_neighbors, sigma, n_components_target)[source]

Locally preserving projection.

Parameters:

  • n_components_target (int)

  • sigma (float)

  • descriptor_data (float64) – 2D array (typically, numpy.array)

  • graph_nearest_neighbors (int) – number (knn)

Returns:

AtomicAI.dim_reduction.lpp.lpp(features, reduced_dim, sigma)[source]
AtomicAI.dim_reduction.lpp.new_centering(reference_data, my_data)[source]

Keep only columns that have some variance.

Parameters:

  • reference_data

  • my_data

Returns:

AtomicAI.dim_reduction.lpp.perform_lpp(descriptor_data, graph_nearest_neighbors=None, sigma=None, number_of_dimensions=None)[source]

Note

dim_reduction and dim_reduction_mpi are available as command-line tools. See Dimensionality Reduction for usage details.