Inside-Out Cross-Covariance Models for Spatial Multivariate Data

A methodological dataset and R package for spatially aware likelihood-based inference on multivariate data. The work develops inside-out cross-covariance models, which are scalable and flexible alternatives to the linear model of coregionalization, and demonstrates performance on synthetic data and colorectal cancer proteomics data. The dataset was authored by Michele Peruzzi and last updated on June 4, 2026.

Use Cases

• Developing scalable covariance models for noisy vector data based on the described inside-out cross-covariance method.
• Applying flexible dimension reduction for large spatial datasets as mentioned in the description.
• Benchmarking novel spatially aware methods against the linear model of coregionalization using the provided synthetic data.
• Analyzing spatial dependence in biomedical data, such as the colorectal cancer proteomics example included.

Strengths

• Includes an R package (spiox) implementing the proposed methods, facilitating reproducibility.
• The 77.9 MB size suggests the dataset contains substantial supporting materials or code.
• Methods are validated on both synthetic data and a real-world application (colorectal cancer proteomics).

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The description is methodological; the actual data content and structure require inspection.

Provenance

Source

figshare

Collection Method

Likely generated for methodological research and demonstration, as described in the accompanying paper.

Freshness

Last updated 2026-06-04 15:37:56