{ "id": 357091, "title": "Tips & tricks on feature selection from Grandmaster", "url": "/competitions/tabular-playground-series-oct-2022/discussion/357091", "author_name": "", "post_date": "2022-10-03T07:11:45.038418200Z", "votes": 15, "comment_count": 2, "views": 0, "content": "

Feature Engineering techniques:

\n

Below are a few notes on how to engineer new features Feature Engineering Techniques by Chris Deotte of Chris Deotte in IEEE-CIS Fraud Detection competition

\n

Feature Selection Method:
\nFeature engineering and feature selection work is really important (especially for tree base models like lightgbm, xgboost or catboost). Here is some feature selection method you can try:

\n

1.Feature Selection with Null Importances by olivier

\n

2.LOFO Feature Importance by Ahmet Erdem

\n

3.Permutation importance

\n

4.Recursive Feature Elimination by ELI5.

\n

5.Feature selection using the Boruta-SHAP package

\n

6.Shapley values: Advanced Uses of SHAP Values and Shapley Values for Feature selection ASHRAE

\n

Reference:

\n

1.Some Feature Selection Technique by KhanhVD
\n2.Feature Engineering Techniques by Chris Deotte

", "messages": [ { "id": "1968688", "postDate": "10/03/2022 07:11:45", "content": "

Feature Engineering techniques:

\n

Below are a few notes on how to engineer new features Feature Engineering Techniques by Chris Deotte of Chris Deotte in IEEE-CIS Fraud Detection competition

\n

Feature Selection Method:
\nFeature engineering and feature selection work is really important (especially for tree base models like lightgbm, xgboost or catboost). Here is some feature selection method you can try:

\n

1.Feature Selection with Null Importances by olivier

\n

2.LOFO Feature Importance by Ahmet Erdem

\n

3.Permutation importance

\n

4.Recursive Feature Elimination by ELI5.

\n

5.Feature selection using the Boruta-SHAP package

\n

6.Shapley values: Advanced Uses of SHAP Values and Shapley Values for Feature selection ASHRAE

\n

Reference:

\n

1.Some Feature Selection Technique by KhanhVD
\n2.Feature Engineering Techniques by Chris Deotte

", "rawMarkdown": "Feature Engineering techniques:\n\nBelow are a few notes on how to engineer new features [Feature Engineering Techniques](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/108575) by Chris Deotte of Chris Deotte in IEEE-CIS Fraud Detection competition\n\nFeature Selection Method:\nFeature engineering and feature selection work is really important (especially for tree base models like lightgbm, xgboost or catboost). Here is some feature selection method you can try:\n\n1.[Feature Selection with Null Importances](https://www.kaggle.com/code/ogrellier/feature-selection-with-null-importances/comments) by olivier\n\n2.[LOFO Feature Importance](https://www.kaggle.com/code/aerdem4/optiver-lofo-feature-importance/notebook) by Ahmet Erdem\n\n3.[Permutation importance](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/107877)\n\n4.[Recursive Feature Elimination](https://www.kaggle.com/code/nroman/recursive-feature-elimination) by ELI5.\n\n5.[Feature selection using the Boruta-SHAP package](https://www.kaggle.com/code/carlmcbrideellis/feature-selection-using-the-boruta-shap-package/notebook)\n\n6.Shapley values: [Advanced Uses of SHAP Values](https://www.kaggle.com/code/dansbecker/advanced-uses-of-shap-values/tutorial) and [Shapley Values for Feature selection ASHRAE](https://www.kaggle.com/code/hmendonca/shapley-values-for-feature-selection-ashrae)\n\nReference:\n\n1.[Some Feature Selection Technique](https://www.kaggle.com/competitions/optiver-realized-volatility-prediction/discussion/269283#1497225) by KhanhVD\n2.[Feature Engineering Techniques](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/108575) by Chris Deotte", "votes": null }, { "id": "1968883", "postDate": "10/03/2022 08:53:29", "content": "

Point 1 is great @validmodel

", "rawMarkdown": "Point 1 is great @validmodel", "votes": null }, { "id": "1969836", "postDate": "10/03/2022 17:53:42", "content": "

You can also include the Attention mechanism for model interpretability.

", "rawMarkdown": "You can also include the Attention mechanism for model interpretability.", "votes": null } ], "comments": [ { "id": 1968883, "author_name": "abrafey", "author_url": "", "post_date": "10/03/2022 08:53:29", "content": "

Point 1 is great @validmodel

", "votes": null, "replies": [] }, { "id": 1969836, "author_name": "debashis74017", "author_url": "", "post_date": "10/03/2022 17:53:42", "content": "

You can also include the Attention mechanism for model interpretability.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1968688": "Feature Engineering techniques:\n\nBelow are a few notes on how to engineer new features [Feature Engineering Techniques](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/108575) by Chris Deotte of Chris Deotte in IEEE-CIS Fraud Detection competition\n\nFeature Selection Method:\nFeature engineering and feature selection work is really important (especially for tree base models like lightgbm, xgboost or catboost). Here is some feature selection method you can try:\n\n1.[Feature Selection with Null Importances](https://www.kaggle.com/code/ogrellier/feature-selection-with-null-importances/comments) by olivier\n\n2.[LOFO Feature Importance](https://www.kaggle.com/code/aerdem4/optiver-lofo-feature-importance/notebook) by Ahmet Erdem\n\n3.[Permutation importance](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/107877)\n\n4.[Recursive Feature Elimination](https://www.kaggle.com/code/nroman/recursive-feature-elimination) by ELI5.\n\n5.[Feature selection using the Boruta-SHAP package](https://www.kaggle.com/code/carlmcbrideellis/feature-selection-using-the-boruta-shap-package/notebook)\n\n6.Shapley values: [Advanced Uses of SHAP Values](https://www.kaggle.com/code/dansbecker/advanced-uses-of-shap-values/tutorial) and [Shapley Values for Feature selection ASHRAE](https://www.kaggle.com/code/hmendonca/shapley-values-for-feature-selection-ashrae)\n\nReference:\n\n1.[Some Feature Selection Technique](https://www.kaggle.com/competitions/optiver-realized-volatility-prediction/discussion/269283#1497225) by KhanhVD\n2.[Feature Engineering Techniques](https://www.kaggle.com/competitions/ieee-fraud-detection/discussion/108575) by Chris Deotte", "1968883": "Point 1 is great @validmodel", "1969836": "You can also include the Attention mechanism for model interpretability." }, "source": "meta" }