Suggestion for functionality

[mileta1976]

Hello Bernado, your package certainly has great potential !

Could you add this functionality: https://towardsdatascience.com/my-secret-sauce-to-be-in-top-2-of-a-kaggle-competition-57cff0677d3c

Thank you !

[laresbernardo]

Hi @mileta1976, thanks!! I don’t know exactly what are you looking for but have you checked this post? https://datascienceplus.com/understanding-titanic-dataset-with-h2os-automl-dalex-and-lares-library/ You could also use distr function to check the relationship between 2 variables: https://laresbernardo.github.io/lares/reference/distr.html And corr_var or lasso_var to find most relevant features.

[mileta1976]

Hi Bernarno,

Thank you for quick response.

I know for DALEX and dist function, but we can’t identify noisy features using feature importance. (because they could be fairly important and still be very noisy).

Functionality: Identify noisy features by comparing feature trends in train/validation to identify noisy ones. More info: https://github.com/abhayspawar/featexp/blob/master/featexp_demo.ipynb Thank you !

[NikitaEGroot]

Hi, thanks for your amazing work! I might be missing something obvious, but is there any way to add titles and remove the "Ranked Cross-Correlations" heading from the corr_cross?

[laresbernardo]

Hi @NikitaEGroot thanks! Yes, you can actually do something like:

... + ggplot2::ggtitle("This is my new title")

[NikitaEGroot]

Hi @NikitaEGroot thanks! Yes, you can actually do something like:

... + ggplot2::ggtitle("This is my new title")

Thank you so much! I didn't realize the whole plot worked with ggplot2, this is perfect!