k-means, hierarchical clustering, DBSCAN, PCA, t-SNE, UMAP, Gaussian mixtures and cluster evaluation.
40 cards · basic cards · AI-written, checked twice. Edit anything.
- What does k-means minimize during optimization?
Show answer
The sum of squared distances between points and their assigned cluster centroids- How are cluster centroids updated in k-means?
Show answer
By computing the mean of all points currently assigned to that cluster- What is the elbow method used for in k-means?
Show answer
Selecting the optimal k by plotting inertia versus k and identifying the point where inertia stops decreasing sharply- What is k-means++ initialization?
Show answer
A seeding strategy that spreads initial centroids probabilistically to improve convergence and avoid poor local optima- Name one major limitation of k-means.
Show answer
It assumes clusters are convex and roughly equal-sized, and fails on elongated or nested clusters- What is the core difference between agglomerative and divisive hierarchical clustering?
Show answer
Agglomerative starts with n clusters (one per point) and merges upward; divisive starts with one cluster and splits downward- What does a dendrogram represent in hierarchical clustering?
Show answer
A tree diagram showing the sequence of cluster merges or splits and the distance at which each occurs- Define complete linkage in hierarchical clustering.
Show answer
The distance between two clusters is the maximum distance between any pair of points from different clusters- What is Ward linkage in hierarchical clustering?
Show answer
A linkage method that merges clusters to minimize the increase in within-cluster variance- How does DBSCAN handle outliers?
Show answer
Points that are neither core points nor density-reachable from any core point are labeled as noise (outliers)- What is a computational challenge with DBSCAN?
Show answer
Finding neighbors requires distance computations that scale poorly with dataset size and dimensionality- What is the primary objective of PCA?
Show answer
To reduce dimensionality by finding orthogonal directions that maximize variance in the data- How much of the original data can be reconstructed from principal components?
Show answer
All variance can be recovered if all components are retained; typically some variance is lost when components are discarded- Why must principal components be orthogonal in PCA?
Show answer
Orthogonality ensures components are uncorrelated, avoiding redundant information and providing a unique basis- What does the explained variance ratio tell us in PCA?
Show answer
The fraction of total variance in the data captured by each principal component, useful for deciding how many components to keep
FlashKeepers