Projector view

View the data rows associated with each model run as data points in a cloud visualization.

When you select a model from the Model tab, you will have three views to choose from, the gallery view, the metrics view, and the projector view.

To switch to the projector view, click on the projector icon in the top right corner.

3456

The projector view helps you explore and understand the data of a model run

Follow these steps to access the projector view in the Model tab.

  1. Go to the Model tab.

  2. Navigate to the model and model run you want to explore.

  3. Click on the icon of the projector view (see screenshot above).

Embedding projector

Embedding projector is a tool for uncovering patterns in unstructured data that can be used to diagnose systemic model and labeling errors. The embedding projector works for data rows that have embeddings. Most embeddings have much higher dimensions than 2 or 3, making it impossible to identify global patterns. You can use the projector tool to employ dimensionality reduction algorithms to explore embeddings in 2D interactively. You can navigate to the embedding projector by using the icons in the top right.

3456

We support two algorithms for dimensionality reduction, UMAP & PCA, which can be toggled. PCA is much faster but does not attempt to find clusters in the data. UMAP is better for trying to find regions of similar points in the data. Sphereize data normalizes the data by subtracting the mean and dividing by the norm.

To learn more about UMAP and dimensionality reduction, check out this guide below from Google.

Supported methods

MethodSpeedClustering
PCA: Principal Component AnalysisFastNo
UMAP: Uniform Manifold Approximation and ProjectionSlowYes

Use the selection tool to isolate data rows for further investigation. To drill into a set of data rows further, you can click the selected data rows button in the top right and click Filter to select. This will re-run the dimensionality reduction algorithm you selected on the subset of data rows to surface local clusters in the data.


What’s Next