<p>Comparison of five dimensionality reduction methods across local and global structure p

5.5 KB of evaluation results for five dimensionality reduction methods, created by Uma Shashi Sharma and last updated in June 2026. The dataset compares methods using combined local and global structure preservation scores, as well as biologically relevant relationships based on expert annotations. It includes a final column reporting average overlap across expert-label class pairs.

Use Cases

• Benchmarking dimensionality reduction algorithms based on local structure (LS) and global structure (GS) scores.
• Evaluating method performance for biological data analysis based on expert-annotated relationships and class separation (AO).
• Comparing the impact of Euclidean versus Cosine distance on preserving biologically relevant structures (BTS).
• Selecting a dimensionality reduction method by identifying the best and second-best performers for each metric.

Strengths

• Provides a direct comparison of five methods using multiple established metrics (T(k), LCMC, Pearson's, Spearman's).
• Includes expert annotations to ground the evaluation in biologically relevant relationships.
• Results are clearly marked, with bold and underlined values indicating the top two performing methods for each metric.
• Released under a permissive CC-BY-4.0 license for open reuse.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small (5.5 KB), indicating limited scope.

Provenance

Source

figshare

Collection Method

Likely contains computed scores from a methodological comparison study.

Freshness

Last updated 2026-06-02 17:40:32; freshness should be verified.