An empirical benchmark ranks 36 dataset similarity methods for continuous data and recommends decision rules plus small combinations that match top performance in 90-95% of tested two- and multi-sample scenarios.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
stat.ME 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
An Empirical Comparison of Methods for Quantifying the Similarity of Numeric Datasets
An empirical benchmark ranks 36 dataset similarity methods for continuous data and recommends decision rules plus small combinations that match top performance in 90-95% of tested two- and multi-sample scenarios.