Inferences on Mixing Probabilities and Ranking in Mixed-Membership Models

A 9.8 MB collection of research materials on the Degree-Corrected Mixed Membership (DCMM) model for network data. The dataset, authored by Sohom Bhattacharya and last updated in May 2026, includes PDF and ZIP files containing theoretical derivations, a ranking scheme for vertices, and numerical experiments on real and synthetic data. It provides methods for asymptotic distributions, confidence intervals, and a multiplier bootstrap for ranking inference of individual membership profiles.

Use Cases

• Quantifying uncertainty in node community membership based on derived finite-sample expansions.
• Ranking vertices within a network according to their membership mixing probabilities on specific communities.
• Performing statistical inference on population quantities like membership vectors using the proposed asymptotic distributions.
• Validating mixed-membership model methods through the provided numerical experiments on synthetic and real data examples.

Strengths

• Includes a novel finite-sample expansion for membership mixing probabilities, filling a gap in uncertainty quantification.
• Demonstrates validity via numerical experiments on both real and synthetic data.
• Released under a permissive CC-BY-4.0 license, allowing for broad reuse and modification.

Limitations

• Row count and column-level documentation are absent, requiring manual inspection after download to understand data structure.
• The dataset appears to be primarily research output (PDF/ZIP) rather than a raw, analysis-ready data table.

Provenance

Source

figshare

Collection Method

Likely contains research outputs including theoretical derivations and experimental results.

Freshness

Last updated 2026-05-18 14:16:32