Source Of Summary:

Lec.4.D. M. FCDS spring 2024.pdf

Data Mining and Analytics – Lecture 4

1. Overview of Clustering Methods 🎯

Clustering is used to group similar data objects together. In this lecture, we explore several clustering methods:

• Partitioning Methods:
  • Examples: k-means & k-medoids
  • 🔹 Objects are divided into non-overlapping clusters.
• Hierarchical Clustering:
  • Builds a tree-like structure (dendrogram) of clusters 🌳.
• Fuzzy Clustering:
  • Clusters may overlap; an object can belong to more than one cluster 🤹‍♀️.
• Density-Based Clustering:
  • Groups data based on density (e.g., DBSCAN) 🌌.
• Model-Based Clustering:
  • Assumes data are generated by underlying probability distributions 📊.

2. Hard vs. Soft (Fuzzy) Clustering 🔄

• Hard Clustering:
  • Each document or data object belongs to exactly one cluster. 🚫🔀
  • Example: A document is either in Cluster A or not.
• Soft (Fuzzy) Clustering:
  • Clusters can overlap; an object may belong to multiple clusters with varying degrees of membership. 🌈
  • Example: A document could be 70% in Cluster A and 30% in Cluster B.

3. Introduction to Fuzzy Logic 🤔💡

Fuzzy Logic is an approach that uses "degrees of truth" rather than the strict true/false (1 or 0) values of Boolean logic.

Key points: