Cubert Interactive 3D visualization of datasets

  • image1

    Adult

  • image2

    Covertype - big

  • image1

    Covertype - small

  • image1

    IJCNN

  • image1

    MNIST

  • image1

    News 20

  • image1

    Cadata

  • image1

    CPU usage

Click the cube above to open an example visualization in a new window. It takes a few seconds to load.

Cubert allows you to visualize your data in 3D, as described in Interactive in-browser 3D visualization of datasets. The method we use to reduce dimensionality is PCA.

You can upload a file in CSV or Libsvm format (classification only, either binary or multiclass) and get a link to a visualization.

The optimal number of examples in a file is roughly 10 thousand to 500 thousand. Please try small files first. That way visualizations will load faster in your browser.

For CSV files, we expect the label in the first column. All data needs to be numeric.