• when using UMAP dimension reduction, you may find weird clusters of data
  • if you remove rows where they have a specific value for a feature, then re-run the dimension reduction, the cluster may disappear
    • this can tell you which feat was causing the clusters
  • To be sure that dropping those rows remove the feature, try 100 seeds and verify that the cluster is gone