Multivariate Embedding

Map High Dimensional Data to a Low Dimensional Space

The Multivariate Embedding platform is available only in JMP Pro.

The Multivariate Embedding platform enables you to map data from very high-dimensional spaces to a low-dimensional space. Many times, you want to map the data to either two or three dimensions so that the low-dimensional space can be easily visualized. In the Multivariate Embedding platform, you can use either the Uniform Manifold Approximation and Projection (UMAP) method or the t-Distributed Stochastic Neighbor Embedding (t-SNE) method. Both methods attempt to fill the low-dimensional space in such a way that clusters of near neighbors can be more easily identified.

Figure 11.1 Example of Multivariate Embedding 

Contents

Overview of the Multivariate Embedding Platform

Example of Multivariate Embedding

Launch the Multivariate Embedding Platform

The Multivariate Embedding Report

Multivariate Embedding Platform Options

Additional Example of Multivariate Embedding

Statistical Details for the Multivariate Embedding Platform

Statistical Details for the t-SNE Method

Statistical Details for the Gradient Descent Algorithm