Computer Science
3 years ago

New test improves accuracy of high-dimensional data analysis in research.

2 views
Journal of nonparametric statistics
A revisit to Bai–Saranadasa's two-sample test
Jin‐Ting Zhang, Tianming Zhu

Paper Summary

The BS-test for high-dimensional data may not control size well if conditions aren't met. A new normal-reference test is proposed, approximating the null distribution of the BS-test using a chi-square-type mixture. This new test performs better in size control than the BS-test and other competitors in simulation studies and real data examples.

Related papers

Theme background image
Mathematics
2 years ago

New Test Improves Accuracy of Data Analysis for Censored Information.

Skew Distributions
Missing Data
Experimental Design
1 view
Theme background image
Computer Science
3 years ago

New test improves accuracy of high-dimensional data analysis in research.

Model-Based Clustering
Skew Distributions
Missing Data
2 views
Theme background image
Computer Science
7 years ago

Discrete data analysis: U-test outperforms t-test in robustness and power.

Big Data Analysis
Health Care Research
Fashion Trends