VIASCKDE Index: A Novel Internal Cluster Validity Index for Arbitrary-Shaped Clusters Based on the Kernel Density...

Area of Science:

• Machine Learning
• Data Mining
• Cluster Analysis

Background:

• Cluster evaluation is crucial in machine learning and data mining.
• Existing cluster validity indices often fail with non-spherical cluster shapes.
• Accurate cluster quality assessment remains a significant challenge.

Purpose of the Study:

• To introduce a novel cluster validity index capable of evaluating arbitrary-shaped clusters.
• To address the limitations of current indices that primarily work for spherical clusters.
• To enhance the accuracy of cluster quality measurement in real-world datasets.

Main Methods:

• Proposed the Validity Index for Arbitrary-Shaped Clusters based on kernel density estimation (VIASCKDE).
• Utilized data separation and compactness metrics to support arbitrary cluster shapes.
• Employed kernel density estimation (KDE) to prioritize denser regions within clusters for improved compactness.

Main Results:

• The VIASCKDE Index demonstrated superior performance compared to state-of-the-art cluster validity indices.
• Experimental results validated the effectiveness of the VIASCKDE Index in evaluating non-spherical clusters.
• The index accurately measures clustering quality by considering both separation and compactness.

Conclusions:

• The VIASCKDE Index offers a more accurate and robust solution for cluster evaluation, especially for arbitrary-shaped clusters.
• This new index advances the field of cluster analysis by overcoming limitations of traditional methods.
• The findings suggest VIASCKDE as a valuable tool for machine learning and data mining applications.