"not selecting any important features since it did not meet criteria: F-score > 1.0"

[EXJUSTICE]

Hi,

I've a dataset with 104 features and over 90k rows for a numerical regression problem, and am working in Kaggle, using the specific installation instructions mentioned. I've bumped into the problem that FeatureWiz is claiming that none of my features match the F-score limit required. However, running the XGB model with features definitely results in a capability to achieve an RMSE baseline. From F-Score charts, I can observe that the maximal F-score is 0.0125 for one of the XGB models.

Any help is appreciated with this, wondering if I would have to change the F-score limit somehow for this case?

[AutoViML]

Hi @EXJUSTICE : thank you for trying out featurewiz.

I don't know where the F-score limit is reached message is coming from. I need a screenshot or at least a Kaggle notebook where you are seeing that message.

Can you please add more details to help me? Thanks AutoVimal

[EXJUSTICE]

I'll get back to you when the competition's over in a few days with a more detailed breakdown of the problem.

[AutoViML]

Can you post the Kaggle data set or problem that you are working on. I can look it up myself.

Auto Vimal

[AutoViML]

Hi @EXJUSTICE I have found the problem of F-score not selecting any features and fixed it. It happened in some cases. Now it should be fixed.

Please test it with the new version 0.2.5

Thanks👍 AutoVimal

[AutoViML]

HiI am wondering if you had a chance to try featurewiz again. I found the problem with F-score and fixed it. Can you try again?ThanksAuto Vimal On Monday, March 6, 2023 at 03:51:17 AM EST, Yijie Xu @.***> wrote:

I'll get back to you when the competition's over in a few days with a more detailed breakdown of the problem.

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>