Features Selected by SULOV depend on the version of featurewiz about featurewiz HOT 3 CLOSED

Hi @epris 👍
The problem with the second time is seen in your above issue:

SULOV method is erroring. Continuing ...

Since SULOV method is erroring, it gives a different result. Can you please upgrade to the latest version via:

pip install featurewiz --upgrade

Then try again and see if it is indeed that different. When there are 3000+ features while there are only 38 rows, you can expect some variance in results. Just FYI.

Autovimal

from featurewiz.

Hi,
thanks for your response.
Well, I know that 38 samples are few but given the same input, given the same seed, if the algorithm is the same the output is the same.
But, as you noticed, there is a problem with the installation.
Unfortunately, I run the command you suggested but the upgrade get stuck at:
INFO: pip is looking at multiple versions of cloudpickle to determine which version is compatible with other requirements. This could take a while.
Collecting cloudpickle>=1.1.1
Using cached cloudpickle-2.0.0-py3-none-any.whl (25 kB)
Downloading cloudpickle-1.6.0-py3-none-any.whl (23 kB)
Using cached cloudpickle-1.5.0-py3-none-any.whl (22 kB)
Using cached cloudpickle-1.4.1-py3-none-any.whl (26 kB)
Using cached cloudpickle-1.4.0-py3-none-any.whl (25 kB)
Using cached cloudpickle-1.3.0-py2.py3-none-any.whl (26 kB)
Using cached cloudpickle-1.2.2-py2.py3-none-any.whl (25 kB)
Using cached cloudpickle-1.2.1-py2.py3-none-any.whl (25 kB)
Using cached cloudpickle-1.2.0-py2.py3-none-any.whl (24 kB)
Using cached cloudpickle-1.1.1-py2.py3-none-any.whl (17 kB)
INFO: pip is looking at multiple versions of cloudpickle to determine which version is compatible with other requirements. This could take a while.
INFO: This is taking longer than usual. You might need to provide the dependency resolver with stricter constraints to reduce runtime. If you want to abort this run, you can press Ctrl + C to do so. To improve how pip performs, tell us what happened here: https://pip.pypa.io/surveys/backtracking

from featurewiz.

Hi @epris 👍
Thanks for trying out Featurewiz. Here is a Colab notebook which shows how to install featurewiz by carefully following these steps one by one:

!pip install featurewiz --ignore-installed --no-deps
!pip install 'fsspec>=0.3.3'
!pip install category_encoders --upgrade
!pip install imbalanced_ensemble 

https://colab.research.google.com/drive/1ekBZgCDgRGJU25unwZ0lvla_k20pq36h?usp=sharing

from featurewiz.