Structured Conformal Inference for Matrix Completion with Applications to Group Recommende

A methodological paper and associated resources for constructing joint prediction regions for groups of missing entries in a sparsely observed matrix. The work was authored by Ziyi Liang and last updated on 2026-04-28. It includes synthetic experiments and applications to MovieLens datasets to demonstrate the proposed conformal inference method.

Use Cases

• Quantifying uncertainty in group-level recommendations based on the described method for joint prediction regions.
• Benchmarking new matrix completion algorithms against the provided synthetic experiments under various missing-data mechanisms.
• Implementing weighted conformalization frameworks for non-exchangeable data structures as described in the paper.

Strengths

• Includes a 4.1 MB package of resources (PDF, ZIP) related to the method.
• Released under a permissive CC-BY-4.0 license, allowing for reuse and modification.
• Method is described as making no distributional assumptions and providing rigorous inferences by modeling the missingness mechanism.

Limitations

• The actual dataset content, including column definitions and row counts, is unknown and must be inspected after download.
• The description is methodological; the scale and specifics of any accompanying data are not detailed.

Provenance

Source

Ziyi Liang via figshare

Collection Method

Likely includes synthetic data generation and application to public MovieLens datasets, as mentioned in the description.

Freshness

Last updated 2026-04-28 15:15:49