Dataset profile

Benchmark Composition

Aggregate statistics for 198 unique datasets. Individual dataset and suite identities are intentionally not published on this site.

Total datasets

198

100% of the benchmark

Univariate

128

64.6% of datasets

Multivariate

70

35.4% of datasets

Evaluation tracks

4

Protocol-separated reporting

Modality distribution

128 univariate · 70 multivariate

198 unique datasets

Univariate128 · 64.6%
Multivariate70 · 35.4%

Evaluation principles

How aggregate reporting works

Equal weighting

Every dataset contributes equally to aggregate accuracy, regardless of sample count or sequence length.

Reviewed metadata

Modality, domain, size, class count, sequence length, channels, and imbalance are reviewed for all 198 datasets.

Modality views

Overall, Univariate, and Multivariate rankings are computed as distinct, auditable views.

Primary metric

Accuracy is the primary metric; Average Rank and Coverage provide complementary context.

Explore aggregate results

Compare Overall, Univariate, and Multivariate Standard rankings.

Open Leaderboard Evaluation Protocol