MDS High-dimensional Data Dimensionality Reduction Layout

Overview

MDS (Multidimensional Scaling) is a classic dimensionality reduction algorithm. In G6, the MDS layout constructs a distance matrix between nodes and restores their relative distances in high-dimensional space as much as possible in 2D space. It is suitable for graph visualization scenarios that show similarity, distance, or structural relationships between nodes.

Use Cases

• Data dimensionality reduction visualization
• Displaying distance relationships between nodes

Configuration

const graph = new Graph({
  layout: {
    type: 'mds',
    center: [300, 300],
    linkDistance: 100,
  },
});

Options

center

The center coordinates of the layout. All nodes will be symmetrically distributed around this point.

linkDistance

number Default: 50

The ideal distance between nodes. The larger the value, the more dispersed the nodes.

Code Example

createGraph(
  {
    autoFit: 'view',
    data: {
      nodes: Array.from({ length: 25 }, (_, i) => ({
        id: `node-${i}`,
        data: {
          value: Math.random() * 100,
        },
      })),
      edges: Array.from({ length: 20 }, (_, i) => ({
        id: `edge-${i}`,
        source: `node-${Math.floor(Math.random() * 25)}`,
        target: `node-${Math.floor(Math.random() * 25)}`,
      })),
    },
    node: {
      style: {
        labelText: (d) => d.id,
      },
    },
    layout: {
      type: 'mds',
      nodeSize: 32,
      linkDistance: 100,
    },
    behaviors: ['drag-element', 'drag-canvas', 'zoom-canvas'],
  },
  { width: 600, height: 300 },
);