Statistical Methods and Applications

Synthetic Data Generation and Random Projection Methods for Assessing Multivariate Similarity: A Low-Dimensional Methodological Study

This paper evaluates synthetic data generation methods using random projection and multivariate similarity assessment in low-dimensional settings to address small-sample challenges in biomedical research.

The Rayleigh Quotient and Contrastive Principal Component Analysis II

This paper extends contrastive PCA by introducing spatial and functional variants, applying them to analyze gene expression in cancer and immune response to vaccination.

The Rayleigh Quotient and Principal Component Analysis I

This paper introduces rho PCA, a method that improves upon contrastive PCA by approximating the Rayleigh quotient, offering better accuracy and efficiency for dimension reduction in genomics data.

Modelling punctuated similarity

This paper introduces a new similarity model called punctuated similarity to identify critical time points in continuous data, demonstrating its application to child growth and stock market volatility.

An ensemble learning method for joint kernel association testing and principal component analysis on multiple kernels

This paper introduces enKern, an ensemble learning method that jointly performs kernel association testing and principal component analysis across multiple kernels, demonstrating improved validity and power in microbiome beta-diversity analysis.

More or fewer latent variables in the high-dimensional data space? That is the question

This study evaluates methods for determining the optimal number of latent variables in high-dimensional neural data, finding that parallel analysis and cross-validation are most effective.

Projection Statistics ProST Online statistical assessment of group separation in data projection analysis

ProST provides a user-friendly tool for assessing the statistical significance of group separation in data projection analysis using PCA, t-SNE, UMAP, and LDA.

A multivariate outcome test of covariance

The paper introduces MANOCCA, a method to test the effect of predictors on the covariance matrix of multivariate outcomes, showing it outperforms existing methods in simulations and identifies significant associations in real data.

Nonmetric ANOVA: a generic framework for analysis of variance on dissimilarity measures

The paper introduces nonmetric ANOVA (nmA) to perform ANOVA on nonmetric dissimilarity measures, providing a statistical framework for analyzing data where metric properties are not satisfied.

Pairwise Relative Distance (PRED) is an intuitive and robust metric for assessing vector similarity and class separability

The paper introduces PRED, a new vector similarity metric that combines pattern and magnitude comparisons, showing it is robust, invariant to scaling, and effective for class separability.

Association Plots: Visualizing associations in high-dimensional correspondence analysis biplots

The paper introduces Association Plots to visualize cluster-specific measurements in high-dimensional data using Correspondence Analysis, demonstrating their effectiveness on genomic data.

A generalization of partial least squares regression and correspondence analysis for categorical and mixed data: An application with the ADNI data

The paper introduces PLS-CA-R, a method that generalizes partial least squares regression to handle categorical and mixed data, and applies it to ADNI data.

Discriminating the Influence of Correlated Factors from Multivariate Observations: the Back-to-Back Regression

The paper introduces Back-to-Back regression to disentangle correlated factors from multivariate observations, showing it outperforms existing methods in causal identification.

JACOBI4 software for multivariate analysis of biological data

JACOBI4 automates multidimensional statistical analysis for biological data without requiring programming experience.

The GMD-biplot and its application to microbiome data

The paper introduces the GMD-biplot, a novel visualization method for microbiome data that improves clustering and variable-sample relationships compared to traditional PCoA plots.