---
title: "Major clustering approaches"
subject: "Data Mining"
chapter: 5
key: "semester_7/data_mining:clustering-approaches"
source: https://www.bctengineer.com/q/semester_7/data_mining/clustering-approaches
site: BCT Engineering
---

# Major clustering approaches

Data Mining · Chapter 5: Cluster Analysis

- Asked in: year not given
- Video: [Lec - 22: Clustering in Data Mining Explained | Top Clustering Methods You MUST Know!](https://www.youtube.com/watch?v=c8yiJq1do38) — Gate Smashers
- Answer checked by Claude Opus 5.5 (high effort) on 2026-10-14
- Page: https://www.bctengineer.com/q/semester_7/data_mining/clustering-approaches
- Whole chapter: https://www.bctengineer.com/md/semester_7/data_mining/chapter-5

## Question

Describe major clustering approaches. Differentiate between hierarchical and partitioning clustering technique.

## Short answer

Major approaches: partitioning (e.g., k-means), hierarchical (aglomerative/divisive), density-based (DBSCAN), grid-based (STING), model-based (EM). Hierarchical builds a dendrogram without prespecifying k; partitioning divides data into k clusters iteratively, requires k, is faster, and typically finds spherical clusters.

## Answer

Major clustering approaches are:

1. **Partitioning methods**: divide n objects into k clusters, then improve the partition by moving objects between clusters. Distance-based, find spherical clusters, need k. *Examples*: k-means, k-medoids (PAM), CLARA.
2. **Hierarchical methods**: create a tree (dendrogram) of clusters.
   - *Agglomerative* (bottom-up, AGNES) merges the closest clusters.
   - *Divisive* (top-down, DIANA) splits clusters.
   - Examples: BIRCH, CURE, Chameleon.
3. **Density-based methods**: a cluster is a dense region of objects separated by sparse regions; can find arbitrary shapes and handle noise. *Examples*: DBSCAN, OPTICS, DENCLUE.
4. **Grid-based methods**: quantise the space into a finite grid of cells and cluster the cells; speed independent of the number of objects. *Examples*: STING, CLIQUE.
5. **Model-based methods**: assume a model (e.g. Gaussian mixture) for each cluster and fit the data to it. *Examples*: EM, COBWEB, SOM.

#### Hierarchical vs partitioning
| Basis | Hierarchical | Partitioning |
| --- | --- | --- |
| Output | Nested clusters, dendrogram | Single flat partition |
| Number of clusters | Not required in advance | k must be given |
| Method | Merge (agglomerative) or split (divisive) | Iterative relocation |
| Reversibility | A merge/split cannot be undone | Objects can change clusters |
| Complexity | ≥ O(n²), slow | O(nkt), faster |
| Cluster shape | Can be arbitrary (single link) | Mostly spherical |
| Examples | AGNES, DIANA, BIRCH | k-means, PAM |

## Check yourself

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

1. Which clustering approach defines a cluster as a dense region separated by sparse regions and handles noise?
   A. Partitioning
   B. Density-based (DBSCAN)
   C. Grid-based
   D. Agglomerative
   Answer: B. Density-based methods such as DBSCAN, OPTICS and DENCLUE find arbitrary-shaped dense regions and handle noise. (Key point: Density-based methods)

2. What is the main idea of grid-based methods like STING and CLIQUE?
   A. Fit a Gaussian mixture to the data
   B. Quantise the space into a finite grid of cells and cluster the cells
   C. Merge closest clusters bottom-up
   D. Move objects between k clusters
   Answer: B. Grid methods cluster grid cells, so speed depends on the number of cells, not the number of objects. (Key point: Grid-based methods)

3. Which difference between hierarchical and partitioning clustering is correct?
   A. Hierarchical needs k in advance; partitioning does not
   B. Hierarchical gives a dendrogram and cannot undo merges; partitioning gives a flat result with relocation
   C. Both give flat results of the same speed
   D. Partitioning is always slower, at least O(n^2)
   Answer: B. Hierarchical output is a tree with irreversible steps and no k; partitioning needs k, is faster (O(nkt)) and relocates objects. (Key point: Hierarchical vs partitioning)

4. Which pair are model-based clustering methods?
   A. EM and COBWEB
   B. STING and CLIQUE
   C. AGNES and DIANA
   D. k-means and PAM
   Answer: A. Model-based methods assume a model for each cluster; examples are EM, COBWEB and SOM. (Key point: Model-based methods)

## Related in Data Mining

- [Need of clustering; hierarchical vs partitioning; K-means](https://www.bctengineer.com/question.md?key=semester_7%2Fdata_mining%3Acluster-need-hier-vs-partition-kmeans) — Data Mining, chapter 5
- [Short note: Partitioning and hierarchical clustering](https://www.bctengineer.com/question.md?key=semester_7%2Fdata_mining%3Apartition-hierarchical-clustering-note) — Data Mining, chapter 5
- [Hierarchical clustering: AGNES and DIANA](https://www.bctengineer.com/question.md?key=semester_7%2Fdata_mining%3Aagnes-diana) — Data Mining, chapter 5
- [Short note: Hierarchical clustering](https://www.bctengineer.com/question.md?key=semester_7%2Fdata_mining%3Ahierarchical-clustering-note) — Data Mining, chapter 5
- [Short note: Partitioning clustering](https://www.bctengineer.com/question.md?key=semester_7%2Fdata_mining%3Apartitioning-clustering-note) — Data Mining, chapter 5