PCA

The pca() function performs spectral dimensionality reduction on a DataCube by computing its principal components. It decomposes the spectral signatures into orthogonal axes ordered by variance, then projects each pixel’s spectrum onto the first N components.

Example

The following example demonstrates PCA on a synthetically generated :class:DataCube, extracting the first three principal components.

import wizard
from wizard._processing.cluster import pca
from wizard._utils.example import generate_pattern_stack

data = generate_pattern_stack(50, 200, 200, n_circles=2, n_rects=2, n_triangles=2)
dc = wizard.DataCube(data)

# reduce to 10 components
dc_reduced = pca(dc, n_components=10)
print("Reduced cube shape:", dc_reduced.cube.shape)

Output:

Reduced cube shape: (10, 200, 200)