Statistical Methods and Applications

On stability of Canonical Correlation Analysis and Partial Least Squares with application to brain-behavior associations

The paper evaluates the stability of CCA and PLS in high-dimensional datasets, finding instability in small samples and proposing a power calculator for reliable multivariate analysis.

Phantom oscillations in principal component analysis

This paper identifies phantom oscillations in PCA caused by data smoothness and spatial/temporal shifts, showing how they affect analysis and proposing detection methods.

Similarity-Based Multimodal Regression

This paper introduces a distance-based regression method for multimodal data that can handle different data types and dimensionalities, showing its effectiveness through simulations and real-world studies.

Techniques to Produce and Evaluate Realistic Multivariate Synthetic

The paper evaluates a synthetic data generation technique to address the small sample problem by creating statistically similar synthetic populations.

Variable-Selection ANOVA Simultaneous Component Analysis (VASCA)

VASCA improves upon ASCA by selecting variables to increase statistical power without increasing Type-I error, showing better performance than ASCA and FDR in simulations and real data.

So you think you can PLS-DA?

The paper compares PLS-DA and PCA for feature selection using synthetic and microbiome data, showing PCA can outperform PLS-DA in some cases.

Test-statistic correlation and data-row correlation

The paper derives an analytical formula showing test-statistic correlations are generally weaker than data-row correlations, challenging common approximations in statistical analysis.

Goodness-of-fit filtering in classical metric multidimensional scaling with large datasets

The paper improves MDS goodness-of-fit calculations by introducing new statistics to filter poorly represented observations in large datasets.