New Testing Procedures for Analyzing High-Dimensional Data
Author Information
Author(s): Frömke Cornelia, Hothorn Ludwig A, Kropf Siegfried
Primary Institution: Hannover Medical School
Hypothesis
Can relevance-shifted tests improve the analysis of high-dimensional multivariate data with small sample sizes?
Conclusion
The proposed nonparametric procedures effectively control the familywise error rate while identifying relevant differences in high-dimensional data.
Supporting Evidence
- The new procedures control the familywise error rate under point null hypotheses.
- Simulation studies confirm the effectiveness of the proposed methods.
- The methods are particularly useful for analyzing microarray data.
Takeaway
This study introduces new ways to analyze complex data with many variables, helping researchers find important differences without making mistakes.
Methodology
The study proposes two nonparametric procedures that utilize relevance-shifted tests and include multiplicity corrections for high-dimensional data analysis.
Limitations
The procedures may lose power with increasing sample sizes and heterogeneous variances among variables.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website