← All skills
Skill · Unsupervised learningLive

Clustering

Group data with no labels at all — K-Means, the elbow method, and dimensionality reduction with PCA.

  1. Task 1Your first K-MeansEasyGroup unlabelled points into clusters.~11 min · 1 blanks→
  2. Task 2Assign new pointsEasyRoute fresh data to the cluster it's nearest to.~11 min · 1 blanks→
  3. Task 3Pick K with the elbowMediumUse inertia to reason about how many clusters to use.~13 min · 1 blanks→
  4. Task 4Scale before you clusterMediumStop a big-range feature from hijacking the distances.~13 min · 1 blanks→
  5. Task 5Silhouette scoreMediumPut a number on how clean your clustering is.~13 min · 1 blanks→
  6. Task 6Cluster real dataHardDiscover the species in Iris without ever seeing the labels.~14 min · 1 blanks→
  7. Task 7PCA to 2DHardSquash many features down to two you can plot.~13 min · 1 blanks→
  8. Task 8Variance explainedHardCheck how much information your reduction kept.~13 min · 1 blanks→
  9. Task 9DBSCAN & densityHardFind oddly-shaped clusters K-Means would butcher.~15 min · 1 blanks→
  10. Task 10Cluster + reduceHardReduce first, then cluster — the standard high-dim recipe.~16 min · 1 blanks→