jsvir / lscae Goto Github PK

Hi,

Thanks for the cool work! When I run this code on my own data, I find that the selected features are not unique. In my case, I have around 1600 features, and I try to select a subset of features using your algorithm. However, there are 880 unique features if I set the k_selected paramter to 1500. The duplication is true even if i choose a small subset of features, for example, 200 out of 1600.

My question is, how can I get exact K distinct features if k_selected==K? I want to figure out the minimal number of features needed for my task.

Hello,
I'm trying to run your example on Google Colab, but I'm getting this error when installing from pip.

!pip install lscae
ERROR: Could not find a version that satisfies the requirement lscae (from versions: none)
ERROR: No matching distribution found for lscae

Any suggestion would be appreciated.