Unsupervised Learning in Python: Clustering and Patterns
Unsupervised learning finds structure in unlabeled data. Where supervised learning needs answers to learn from, unsupervised algorithms discover patterns, groupings, and relationships on their own. This makes them essential for exploratory data analysis, customer segmentation, anomaly detection, and dimensionality reduction.
This path covers the major unsupervised techniques: clustering algorithms that group similar data points, dimensionality reduction methods that simplify high-dimensional data, and association rule learning for discovering relationships.
Tutorials marked with the cert badge include a final exam that awards a certificate of completion you can download and share.
Unsupervised Learning Models in Python
Overview of unsupervised learning categories, evaluation metrics, and choosing the right algorithm.
Python Clustering Models
Survey of clustering algorithms with comparisons on accuracy, scalability, and cluster shape handling.
Python K-Means Clustering
K-means algorithm, choosing k with the elbow method, initialization strategies, and limitations.
Python DBSCAN Clustering
Density-based clustering that handles noise and irregular cluster shapes without specifying k.
Python Hierarchical Clustering
Agglomerative clustering, dendrograms, linkage methods, and cutting the tree at the right level.
Python Gaussian Mixture Models
Probabilistic clustering with GMMs, expectation-maximization, and soft cluster assignments.
Python Dimensionality Reduction
Why dimensionality reduction matters, the curse of dimensionality, and choosing the right technique.
Dimensionality Reduction Models
Comparison of PCA, t-SNE, UMAP, LDA, and autoencoders for different use cases.
Python Principal Component Analysis (PCA)
PCA math explained simply, variance retention, choosing components, and scikit-learn implementation.
Python t-SNE
t-SNE for visualization, perplexity tuning, and understanding what t-SNE plots do and don't tell you.
Python Association Rule Learning
Market basket analysis, Apriori algorithm, support/confidence/lift metrics, and mlxtend implementation.