6 Clustering Free Online FlashCards

What is clustering?

Clustering is an unsupervised learning task that groups examples by feature similarity without being given correct group labels in advance.

What does a cluster represent?

A cluster is a pattern produced by chosen features, a distance measure, and an algorithm; it is not necessarily a naturally existing or meaningful category.

Which features could help cluster store customers?

A store could describe customers by visit frequency and average purchase amount, then group customers with similar shopping patterns.

What role does a similarity measure play in clustering?

A similarity or distance measure determines which examples are considered alike. The appropriate measure depends on the features and task.

How does K-means form clusters?

K-means assigns examples to the nearest of a chosen number of centers, updates those centers, and repeats the process.

Why can clustering algorithms group the same data differently?

Different algorithms define clusters using different assumptions, so they can produce different groupings from the same data; no one method suits every dataset.

How does clustering differ from classification?

Classification assigns examples to predefined labels; clustering groups examples by similarity without those labels.

Why might clustering data need scaling?

Features on larger numerical scales can dominate distance calculations, so data may need to be scaled or transformed first.

What are key strengths and limitations of K-means?

K-means is efficient for compact, roughly round clusters, but requires choosing the number of clusters and can be affected by outliers or starting centers.

How does agglomerative hierarchical clustering work?

Agglomerative hierarchical clustering starts with each example alone and repeatedly merges the closest groups; its tree can be cut at different levels.

How does DBSCAN treat dense regions and isolated examples?

DBSCAN groups examples in dense regions and can mark isolated examples as noise. Its results depend on settings such as neighborhood radius and minimum neighborhood size.

What makes Gaussian mixture model assignments probabilistic?

Gaussian mixture models represent data as a mixture of probability distributions and can estimate each example's probability of belonging to each cluster.