---
title: "Short note: Evaluation of clustering"
subject: "Data Mining"
chapter: 5
key: "semester_7/data_mining:clustering-evaluation-note"
source: https://www.bctengineer.com/q/semester_7/data_mining/clustering-evaluation-note
site: BCT Engineering
---

# Short note: Evaluation of clustering

Data Mining · Chapter 5: Cluster Analysis

- Asked in: year not given
- Video: [evaluation of clustering in data mining || data mining || machine learning || cse || ns lectures](https://www.youtube.com/watch?v=hsNOzTq8Ilo) — NS lectures
- Answer checked by Claude Opus 5.5 (high effort) on 2026-10-14
- Page: https://www.bctengineer.com/q/semester_7/data_mining/clustering-evaluation-note
- Whole chapter: https://www.bctengineer.com/md/semester_7/data_mining/chapter-5

## Question

Write a short note on Evaluation of Clustering.

## Short answer

Clustering evaluation uses external measures (purity, Rand index) with known labels, internal measures (SSE, silhouette, Dunn index) without labels, and relative measures (elbow method, silhouette comparison) to compare clusterings. It also assesses clustering tendency (e.g., Hopkins statistic) and determines optimal cluster count via gap statistic or silhouette analysis.

## Answer

**Evaluation of clustering** (cluster validity) assesses whether a clustering is meaningful and how good it is. Since clustering is unsupervised there is no single correct answer, so evaluation is more difficult than for classification. It answers four questions: Is there any non-random structure in the data (**clustering tendency**)? What is the right number of clusters? How well do the clusters fit the data without external information? How do they compare with the known labels or with another clustering?

#### Measures
1. **External (supervised) measures**: compare with known class labels. *Purity, entropy, precision/recall/F-measure, Rand index, adjusted Rand index, normalised mutual information.*
2. **Internal (unsupervised) measures**: use only the data. 
   - **Cohesion** = within-cluster sum of squares (SSE), lower is better.
   - **Separation** = between-cluster sum of squares, higher is better.
   - **Silhouette coefficient** s = (b − a)/max(a, b), where a is the mean distance to objects in the same cluster and b is the mean distance to the nearest other cluster; range −1 to 1.
   - Dunn index, Davies-Bouldin index.
3. **Relative measures**: compare different clusterings (e.g. different k or algorithm). *Example*: the **elbow method** plots SSE against k and picks the k where the curve bends; or choose k with the best silhouette.

#### Other aspects
- **Determining the number of clusters** (elbow, silhouette, gap statistic).
- **Clustering tendency** test (e.g. Hopkins statistic) to check that the data is not uniformly random.
- Evaluation depends on the algorithm: compactness suits k-means, but density-based clusters of arbitrary shape need other indices.

## Check yourself

The answer's key points as multiple-choice questions, with the correct option and why.

1. Why is evaluating clustering harder than evaluating classification?
   A. Clustering is unsupervised, so there is no single correct answer
   B. Clustering has too few attributes
   C. Classification has no measures
   D. Clustering always gives one cluster
   Answer: A. Without class labels there is no ground truth to define correctness. (Key point: Why evaluation is hard)

2. How does the elbow method choose k?
   A. Pick the k with highest SSE
   B. Plot SSE against k and pick where the curve bends
   C. Pick the k equal to number of attributes
   D. Pick k = n
   Answer: B. SSE falls as k increases; the bend (elbow) marks a good k. (Key point: Elbow method)

3. What does a clustering tendency test such as Hopkins statistic check?
   A. Whether the data contain non-random structure rather than being uniformly random
   B. The best linkage
   C. The number of outliers
   D. The class labels
   Answer: A. It checks that there are real clusters before clustering is applied. (Key point: Clustering tendency)

4. Which are internal measures?
   A. Purity and Rand index
   B. Cohesion, separation and silhouette
   C. Precision and recall against labels
   D. Adjusted Rand index
   Answer: B. Internal measures use only the data; the others are external, using known labels. (Key point: Internal measures)