classical-mlunsupervisedclustering
Clustering algorithms
k-means, DBSCAN, hierarchical — unsupervised grouping of data by similarity.
Уровни глубины
L0Intro~2ч
Groups 2D points by eye; runs sklearn KMeans on a toy dataset.
L1Basics~10ч
Implements k-means from scratch; uses elbow / silhouette to pick k; knows DBSCAN's eps.
L2Working~15ч
Chooses between KMeans / GMM / DBSCAN / HDBSCAN / Spectral by data topology; validates with ARI/NMI.
L3Advanced~25ч
Derives EM for GMM; spectral clustering via graph Laplacian; deep clustering.
L4Research~60ч
Contributes to self-supervised clustering, contrastive embeddings for unsupervised discovery.