Dunn Index

INFO

Measures the minimum inter-cluster distance relative to the maximum intra-cluster distance. A higher index value suggests better separation.

How It Works

$$D=\frac{{\min }_{i\ne j}\delta (C_i,C_j)}{{\max }_k\Delta (C_k)}$$

• $\delta (C_i,C_j)$: Distance between clusters $i$ and $j$
• $\Delta (C_k)$: Diameter of cluster $k$

What to Look For

• Higher Dunn Index = better clustering
• Sensitive to noise and outliers
• Best for compact, well-separated clusters

Application Models

• K-Means Clustering
• Hierarchical Clustering
• Density-Based Spatial Clustering of Application with Noise (DBSCAN)
• Gaussian Mixture Models