Feat : k_Medoids Algorithm

[MeKaustubh07]

K-Medoids (PAM) Algorithm – Summary Review

Overview

K-Medoids (Partitioning Around Medoids) clusters data into k groups using actual data points (medoids) as centers — more robust and interpretable than K-Means, especially against outliers.

---

Algorithm Summary

1. Initialize: Randomly choose k medoids
2. Assign: Each point → nearest medoid
3. Update: Replace medoid with point minimizing total dissimilarity
4. Repeat until convergence

Key Traits:

• Uses real points (medoids)
• Works with any distance metric
• Resistant to outliers
• Easy to interpret

---

Strengths ✅

• Well-structured R6 OOP design
• Handles edge cases, validation, and convergence
• Supports multiple distance metrics
• Helpful methods: fit(), predict(), silhouette_score(), get_medoids()
• Excellent documentation and examples
• Includes quality metrics (silhouette, inertia)

---

Improvements 🔧

• Optimize distance matrix (O(n²))
• Add smarter initialization (K-Means++)
• Enable parallel processing
• Add early stopping & multi-run options
• Add visualization (plot_clusters())

---

Complexity

• Time: O(n² × d + i × k × (n-k)²)
• Space: O(n²)
• Scalability: Best for small–medium datasets; use CLARA for large ones

---

K-Medoids vs K-Means

Aspect	K-Medoids	K-Means
Centers	Real data points	Means
Outlier Sensitivity	Low	High
Distance Metrics	Any	Euclidean
Interpretability	High	Moderate
Speed	Slower	Faster

---

Use Cases

✅ Customer segmentation
✅ Document clustering
✅ Noisy or mixed data
❌ Avoid for huge or high-dimensional datasets

---

Testing

• Verify clustering correctness and silhouette range
• Check Euclidean & Manhattan distances
• Edge cases: k=1, identical points, high-dim data

---

Verdict ⭐⭐⭐⭐⭐

Excellent, production-ready implementation — clean, robust, well-documented, and ideal for educational or research use.
Next steps: add optimization, visualization, and CLARA variant for scalability.

[Copilot]

Pull Request Overview

This PR introduces a complete implementation of the K-Medoids (PAM) clustering algorithm in R using R6 object-oriented design. K-Medoids is a robust alternative to K-Means that uses actual data points as cluster centers, making it more resistant to outliers.

Key changes:

• Full K-Medoids implementation with support for Euclidean and Manhattan distance metrics
• Comprehensive API including fit(), predict(), silhouette_score(), and helper methods
• Extensive documentation with roxygen2 comments and four detailed usage examples

[Copilot]

Pull Request Overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

[Copilot]

Pull Request Overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 3 comments.