|Thursday, February 15|
|PS1 Poster Session 1 and Opening Mixer||
Thu, Feb 15, 5:30 PM - 7:00 PM
Appropriate Dimension Reduction for Sparse, High-Dimensional Data Using Intensity Plots and Other Visualizations (303619)
*Eugenie Jackson, West Virginia University
Ekaterina Smirnova, University of Montana
Keywords: ordination methods, dimension reduction, visualization
Dimension reduction for high-dimensional data is necessary for descriptive data analysis. Most visualization options are restricted to 3 dimensions, while more than 3 are necessary to capture the relationships among variables (or observations) in sparse, high-dimensional data. Using 16S rRNA microbiome data, we develop intensity plots to highlight the changing contributions of taxa (or subjects) as the number of principal components of the ordination method is changed. In our intensity plots, intensity indicates the relative position of the taxon/subject within the data, and our `inclusion in intensity deciles' plots provide a quick visualization of taxa/subjects that are close to the `center' or that contribute to dissimilarity. These plots provide a visual way to determine an appropriate amount of dimension reduction. They also allow the exploration of patterns among related subjects or taxa not seen in other visualizations. Supporting visualizations confirm the robustness of the approach and are readily accessible to clients.