Dimensionality reduction: duplicate columns
When running the same algorithm multiple times (usually to try different parameters), either new columns should be created with unique names or the columns created the first time should be re-used. To prevent the table from growing too much, I prefer re-using columns. For this, column names need to be standardized. For PCA, I propose to use PCA.PC1 and PCA.PC2 and for UMAP, UMAP.X1 and UMAP.X2. Upon a click on the "Apply" button, the app should check if they already exist, if not, create them, else populate them with the result of the computation.