I conducted dimension reduction using 2 separate UMAP models (one for numeric data, and one for categorical data). I then combined the 2 models using the "union_intersection_mapper" approach (https://umap-learn.readthedocs.io/en/latest/composing_models.html), to produce the final embedding matrix
When I have a new set of data, how do I use the "transform" function to predict the final embedding matrix? I can use the transform function for the 2 models separately, but I cannot merge the 2 models together (in the new data).
Hi,
I conducted dimension reduction using 2 separate UMAP models (one for numeric data, and one for categorical data). I then combined the 2 models using the "union_intersection_mapper" approach (https://umap-learn.readthedocs.io/en/latest/composing_models.html), to produce the final embedding matrix
When I have a new set of data, how do I use the "transform" function to predict the final embedding matrix? I can use the transform function for the 2 models separately, but I cannot merge the 2 models together (in the new data).
Thanks