{ "id": 359009, "title": "📌 New Feature Distribution Library Featdist ❤️❤️❤️", "url": "/competitions/tabular-playground-series-oct-2022/discussion/359009", "author_name": "", "post_date": "2022-10-10T12:29:42.449031300Z", "votes": 12, "comment_count": 4, "views": 0, "content": "

Hi all,

\n

I am working on a new feature distribution library that is featdist for faster exploratory data analysis in the Kaggle. You can simply compare train and test distributions in one line by using numerical_ttt_dist or categorical_ttt_dist methods. Also, these methods return stats about trend changes and the correlation of trend changes.

\n

You can find an example about featdist in this notebook.

\n

df_stats = numerical_ttt_dist(train=train, test=test, features=num_features, target='team_A_scoring_within_10sec', bin_num=50, ncols=6)
\n\"\"

\n

display(df_stats)
\n\"\"

", "messages": [ { "id": "1980849", "postDate": "10/10/2022 12:29:42", "content": "

Hi all,

\n

I am working on a new feature distribution library that is featdist for faster exploratory data analysis in the Kaggle. You can simply compare train and test distributions in one line by using numerical_ttt_dist or categorical_ttt_dist methods. Also, these methods return stats about trend changes and the correlation of trend changes.

\n

You can find an example about featdist in this notebook.

\n

df_stats = numerical_ttt_dist(train=train, test=test, features=num_features, target='team_A_scoring_within_10sec', bin_num=50, ncols=6)
\n\"\"

\n

display(df_stats)
\n\"\"

", "rawMarkdown": "Hi all,\n\nI am working on a new feature distribution library that is [featdist](https://github.com/Hasan-Basri-Akcay/featdist) for faster exploratory data analysis in the Kaggle. You can simply compare train and test distributions in one line by using numerical_ttt_dist or categorical_ttt_dist methods. Also, these methods return stats about trend changes and the correlation of trend changes.\n\nYou can find an example about featdist in this [notebook](https://www.kaggle.com/code/hasanbasriakcay/tpsoct22-insightful-eda-fe-modeling).\n\n`df_stats = numerical_ttt_dist(train=train, test=test, features=num_features, target='team_A_scoring_within_10sec', bin_num=50, ncols=6)`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3381067%2Fd2a14379e15fb3d2ed0383c4a83260f7%2Fttt.png?generation=1665404614746045&alt=media)\n\n`display(df_stats)`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3381067%2F505ab5e55848577a8e68db7c2601ffc9%2Fdf_stats.png?generation=1665404637811234&alt=media)", "votes": null }, { "id": "1981309", "postDate": "10/10/2022 18:11:36", "content": "

Wow, this is highly useful to get up to speed and analyse data quickly and more effectively. Thanks a lot for sharing @hasanbasriakcay

", "rawMarkdown": "Wow, this is highly useful to get up to speed and analyse data quickly and more effectively. Thanks a lot for sharing @hasanbasriakcay", "votes": null }, { "id": "1981316", "postDate": "10/10/2022 18:14:07", "content": "

Thanks a lot @ravi20076, glad you like it 🙏

", "rawMarkdown": "Thanks a lot @ravi20076, glad you like it 🙏", "votes": null }, { "id": "1982614", "postDate": "10/11/2022 14:19:32", "content": "

Hi, thanks for sharing really good to know about something new, U can also try dataprep library for EDA in a single line of code where it good all kind of plots and distribution

", "rawMarkdown": "Hi, thanks for sharing really good to know about something new, U can also try dataprep library for EDA in a single line of code where it good all kind of plots and distribution", "votes": null }, { "id": "1984685", "postDate": "10/12/2022 19:50:02", "content": "

Thanks for sharing @hasanbasriakcay , useful library.

", "rawMarkdown": "Thanks for sharing @hasanbasriakcay , useful library.", "votes": null } ], "comments": [ { "id": 1981309, "author_name": "ravi20076", "author_url": "", "post_date": "10/10/2022 18:11:36", "content": "

Wow, this is highly useful to get up to speed and analyse data quickly and more effectively. Thanks a lot for sharing @hasanbasriakcay

", "votes": null, "replies": [ { "id": 1981316, "author_name": "hasanbasriakcay", "author_url": "", "post_date": "10/10/2022 18:14:07", "content": "

Thanks a lot @ravi20076, glad you like it 🙏

", "votes": null, "replies": [] } ] }, { "id": 1982614, "author_name": "satyaprakashshukl", "author_url": "", "post_date": "10/11/2022 14:19:32", "content": "

Hi, thanks for sharing really good to know about something new, U can also try dataprep library for EDA in a single line of code where it good all kind of plots and distribution

", "votes": null, "replies": [] }, { "id": 1984685, "author_name": "landfallmotto", "author_url": "", "post_date": "10/12/2022 19:50:02", "content": "

Thanks for sharing @hasanbasriakcay , useful library.

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "1980849": "Hi all,\n\nI am working on a new feature distribution library that is [featdist](https://github.com/Hasan-Basri-Akcay/featdist) for faster exploratory data analysis in the Kaggle. You can simply compare train and test distributions in one line by using numerical_ttt_dist or categorical_ttt_dist methods. Also, these methods return stats about trend changes and the correlation of trend changes.\n\nYou can find an example about featdist in this [notebook](https://www.kaggle.com/code/hasanbasriakcay/tpsoct22-insightful-eda-fe-modeling).\n\n`df_stats = numerical_ttt_dist(train=train, test=test, features=num_features, target='team_A_scoring_within_10sec', bin_num=50, ncols=6)`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3381067%2Fd2a14379e15fb3d2ed0383c4a83260f7%2Fttt.png?generation=1665404614746045&alt=media)\n\n`display(df_stats)`\n![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F3381067%2F505ab5e55848577a8e68db7c2601ffc9%2Fdf_stats.png?generation=1665404637811234&alt=media)", "1981309": "Wow, this is highly useful to get up to speed and analyse data quickly and more effectively. Thanks a lot for sharing @hasanbasriakcay", "1981316": "Thanks a lot @ravi20076, glad you like it 🙏", "1982614": "Hi, thanks for sharing really good to know about something new, U can also try dataprep library for EDA in a single line of code where it good all kind of plots and distribution", "1984685": "Thanks for sharing @hasanbasriakcay , useful library." }, "source": "meta" }