Universally Consistent K-Sample Tests via Dependence Measures
arXiv:1910.08883 · doi:10.1016/j.spl.2024.110278
Abstract
The K-sample testing problem involves determining whether K groups of data points are each drawn from the same distribution. Analysis of variance is arguably the most classical method to test mean differences, along with several recent methods to test distributional differences. In this paper, we demonstrate the existence of a transformation that allows K-sample testing to be carried out using any dependence measure. Consequently, universally consistent K-sample testing can be achieved using a universally consistent dependence measure, such as distance correlation and the Hilbert-Schmidt independence criterion. This enables a wide range of dependence measures to be easily applied to K-sample testing.
6 pages main + 8 pages appendix
References in corpus (15)
- Measuring and testing dependence by correlation of distances
- Equivalence of distance-based and RKHS-based statistics in hypothesis testing
- A consistent multivariate test of association based on ranks of distances
- Distance covariance in metric spaces
- DISCO analysis: A nonparametric extension of analysis of variance
- The Chi-Square Test of Distance Correlation
- From Distance Correlation to Multiscale Graph Correlation
- One-Hot Graph Encoder Embedding
- Discovering and Deciphering Relationships Across Disparate Data Modalities
- The Exact Equivalence of Distance and Kernel Methods for Hypothesis Testing
- Discovering Communication Pattern Shifts in Large-Scale Labeled Networks using Encoder Embedding and Vertex Dynamics
- hyppo: A Multivariate Hypothesis Testing Python Package
- Synergistic Graph Fusion via Encoder Embedding
- Discovering the Signal Subgraph: An Iterative Screening Approach on Graphs
- Learning Interpretable Characteristic Kernels via Decision Forests