Learn/Course 06
Classical ML · weeks 8

Unsupervised learning, anomalies & retrieval

Novel equipment failures appear before reliable labels exist.

Opening production brief

A measurable problem, before a model.

Users

Operators, product owners, and people affected by the unsupervised & retrieval decision.

Data

PCA, explained variance, k-means, DBSCAN with version, owner, event time, and availability contracts.

Constraints

Novel equipment failures appear before reliable labels exist. Every solution must state latency, cost, capacity, and fallback limits.

Success

Offline evidence plus a deployable resolution: A PCA, clustering, anomaly, and nearest-neighbor retrieval investigation.

CHAPTER 1

Frame and baseline

Turn the production problem into explicit data, metric, and baseline contracts.

  1. 01

    PCA and representation geometry

    PCA preserves dominant geometric variation, not automatically the information your decision needs.

    65–90 min · checkpoint · deterministic lab
  2. 02

    Clustering as an operational hypothesis

    A cluster becomes knowledge only when stable geometry connects to a repeatable human or system action.

    65–90 min · checkpoint · deterministic lab
CHAPTER 2

Build and stress

Formalize the machinery, test counterexamples, and expose system limits.

  1. 03

    Density clusters and novel anomalies

    Density methods flag what is locally disconnected; people and outcomes determine what is truly novel and important.

    65–90 min · checkpoint · deterministic lab
  2. 04

    Embeddings and similarity

    Similarity is a modeling choice: the same vectors tell different stories under cosine and dot product.

    65–90 min · checkpoint · deterministic lab
CHAPTER 3

Resolve and operate

Join model behavior to architecture, observability, rollout, and rollback.

  1. 05

    Approximate retrieval under latency

    Approximate retrieval is an explicit recall–latency contract, not a transparent substitute for exact search.

    65–90 min · checkpoint · deterministic lab
  2. HW

    Incident discovery and retrieval system

    Combine PCA, clustering, anomaly triage, embeddings, and ANN search with human review.

    Rubric · staged hints · reference resolution
Prerequisite guidance

2 concepts make this faster.

Prerequisites are guidance, never hard gates.

C03Data & featuresData contracts, point-in-time features, backfills, quality gates, and lineage.C04Classical baselinesA justified baseline across regression, Naive Bayes, kNN, and SVM.
Ready to work the problem?

PCA and representation geometry

Start lesson 01