Dimension Reduction¶
PCA¶
Matrix Decomposition method
project feature to max. variance
SVD¶
A better way to perform PCA that aovid huge computation of \(A^TA\)
sklearn is using SVD algo for PCA function
eigenvector¶
Manifold method
t-SNE¶
t-distributed stochastic neighbor embedding
Manifold method
class sklearn.manifold.TSNE(n_components=2, perplexity=30.0, early_exaggeration=12.0, learning_rate=200.0, n_iter=1000, n_iter_without_progress=300, min_grad_norm=1e-07, metric='euclidean', init='random', verbose=0, random_state=None, method='barnes_hut', angle=0.5, n_jobs=None)
Barnes-Hut t-SNE¶
faster, limited to 2~3 dimension (but enough for visualization)