DBCV: Density-Based Clustering Validation Metric

This collection of synthetic 2D spatial datasets was developed by pajaskowiak for density-based clustering validation (DBCV) and was last updated in May 2023. It provides benchmark distributions specifically designed to evaluate the performance of algorithms such as HDBSCAN and DBSCAN in cluster analysis.

Use Cases

• Benchmarking the DBCV metric against 2D spatial coordinate distributions
• Evaluating HDBSCAN clustering performance on varying density-based clusters
• Testing DBSCAN-clustering-algorithm sensitivity using synthetic 2D point sets

Strengths

• Specifically designed for density-based clustering validation (DBCV) metrics
• Includes 2D spatial distributions tailored for HDBSCAN and DBSCAN testing
• Provides synthetic benchmarks for KDD (Knowledge Discovery in Data) research

Limitations

• Synthetic nature lacks the inherent noise and unpredictability of real-world datasets
• Limited to 2D dimensionality, which may not represent complex high-dimensional clustering challenges
• Specific column headers and total record counts are not explicitly documented in the metadata

Provenance

Source

pajaskowiak via GitHub

Collection Method

synthetic

Freshness

Last updated May 2023; the data is static as it serves as a benchmark for clustering metrics.