Explain dimensionality reduction techniques and when to use them.

Dimensionality reduction simplifies datasets by reducing features while preserving important information and relationships.\n\n• PCA: Linear transformation, preserves variance, good for visualization\n• t-SNE: Non-linear, excellent for visualization, preserves local structure\n• LDA: Supervised, maximizes class separation\n• Feature selection: Filter, wrapper, embedded methods\n\nExample: Customer segmentation with 100 features uses PCA to reduce to 10 components explaining 90% variance, enabling faster clustering and visualization. t-SNE creates 2D plots for stakeholder presentations while maintaining customer group separation.