Dimensionality Reduction

The DimReducer class provides a unified interface for fitting and transforming data with PCA, UMAP, t-SNE, and GTM.

DimReducer

cdr_bench.dr_methods.dimensionality_reduction.DimReducer

Dimensionality reduction class supporting multiple methods.

__init__(params)

fit(X, y=None)

Fit the model.

transform(X)

Transform the data.

fit_transform(X, y=None)

Fit and transform the data.

update_params(**new_params)

Update parameters and reinitialize the model.

check_method_implemented(method) staticmethod

Check if the given method is implemented.

default_params() staticmethod

Returns default parameters for each method.

valid_methods() staticmethod

Returns valid methods for dimensionality reduction.

Usage Example

from src.cdr_bench.optimization.params import UMAPParams
from src.cdr_bench.dr_methods.dimensionality_reduction import DimReducer

params = UMAPParams(n_components=2, n_neighbors=15, min_dist=0.1)
reducer = DimReducer(params)
embedding = reducer.fit_transform(X)