The Home Credit competition was the largest tabular data competition at the time on Kaggle and may yield some valuable insights. Like that competition, and default prediction in the real world, this is an unbalanced classification problem. As in most competitions, validation will be vitally important. Cross-validation may be particularly useful for unbalanced classification problems, but, it could be very time-consuming given the size of the data so sensible choices will need to be made about validation to ensure reasonable accuracy and speed.
\nBelow are some further insights from the previous competition. Not all of the technical and modeling insights may translate to this competition but any domain insights and considerations about the time-series aspect of the competition may be very worthwhile to consider.
\nReflections from a member of the winning team
\n\nSummary of team member's contributions for 2nd place solution
\nSome findings from the 2nd place solution
\n\n\n\n\n9th place solution and experience
\n\nEthical and Discrimination Considerations
\nThe features in the new competition do not fully spell out the meaning of the variables although the first letter of the feature name indicates the domain area where the feature belongs. That could be useful in trying to apply any domain insights.
\nApologies if I missed any of the top solutions. Feel free to add any I missed or any other useful links or recollections from that competition.
", "messages": [ { "id": 1801527, "postDate": "2022-05-25T20:34:08.580Z", "content": "The Home Credit competition was the largest tabular data competition at the time on Kaggle and may yield some valuable insights. Like that competition, and default prediction in the real world, this is an unbalanced classification problem. As in most competitions, validation will be vitally important. Cross-validation may be particularly useful for unbalanced classification problems, but, it could be very time-consuming given the size of the data so sensible choices will need to be made about validation to ensure reasonable accuracy and speed.
\nBelow are some further insights from the previous competition. Not all of the technical and modeling insights may translate to this competition but any domain insights and considerations about the time-series aspect of the competition may be very worthwhile to consider.
\nReflections from a member of the winning team
\n\nSummary of team member's contributions for 2nd place solution
\nSome findings from the 2nd place solution
\n\n\n\n\n9th place solution and experience
\n\nEthical and Discrimination Considerations
\nThe features in the new competition do not fully spell out the meaning of the variables although the first letter of the feature name indicates the domain area where the feature belongs. That could be useful in trying to apply any domain insights.
\nApologies if I missed any of the top solutions. Feel free to add any I missed or any other useful links or recollections from that competition.
", "rawMarkdown": "The Home Credit competition was the largest tabular data competition at the time on Kaggle and may yield some valuable insights. Like that competition, and default prediction in the real world, this is an unbalanced classification problem. As in most competitions, validation will be vitally important. Cross-validation may be particularly useful for unbalanced classification problems, but, it could be very time-consuming given the size of the data so sensible choices will need to be made about validation to ensure reasonable accuracy and speed.\n\nBelow are some further insights from the previous competition. Not all of the technical and modeling insights may translate to this competition but any domain insights and considerations about the time-series aspect of the competition may be very worthwhile to consider. \n\n[Reflections from a member of the winning team](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/58332)\n\n[First place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64821)\n\n[Summary of team member's contributions for 2nd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64722)\n\n[Some findings from the 2nd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64784)\n\n[3rd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64596)\n\n[4th place solution and tips](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64487)\n\n[7th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64580)\n\n[8th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64474)\n\n[9th place solution and experience](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64536)\n\n[10th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64598)\n\n[Ethical and Discrimination Considerations](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/57028)\n\nThe features in the new competition do not fully spell out the meaning of the variables although the first letter of the feature name indicates the domain area where the feature belongs. That could be useful in trying to apply any domain insights.\n\nApologies if I missed any of the top solutions. Feel free to add any I missed or any other useful links or recollections from that competition.", "votes": 93 }, { "id": 1884359, "postDate": "2022-08-04T11:49:43.133Z", "content": "ohh niceπ
", "rawMarkdown": "ohh niceπ", "votes": 1 }, { "id": 1818184, "postDate": "2022-06-12T11:18:10.757Z", "content": "@datahobbit , Thank you so much for considering Ethical considerations here :)
\nThanks
Sure no problem. This is becoming an increasingly important consideration with real-life models, especially those that can make a meaningful difference in the lives of individuals.
", "rawMarkdown": "Sure no problem. This is becoming an increasingly important consideration with real-life models, especially those that can make a meaningful difference in the lives of individuals.", "votes": 1 } ] }, { "id": 1807737, "postDate": "2022-06-01T09:27:32.917Z", "content": "Hi ! Thanks for the source list.βοΈ
", "rawMarkdown": "Hi ! Thanks for the source list.βοΈ", "votes": 1, "replies": [ { "id": 1808202, "postDate": "2022-06-01T15:53:44.780Z", "content": "Happy to help.
", "rawMarkdown": "Happy to help. " } ] }, { "id": 1804737, "postDate": "2022-05-29T11:55:47.077Z", "content": "πππππππππβοΈβοΈ
", "rawMarkdown": "πππππππππβοΈβοΈ", "votes": 1 }, { "id": 1801999, "postDate": "2022-05-26T10:34:23.773Z", "content": "thanks for sharing this @datahobbit , looking to see you at the leaderboard
", "rawMarkdown": "thanks for sharing this @datahobbit , looking to see you at the leaderboard ", "votes": 1, "replies": [ { "id": 1802472, "postDate": "2022-05-26T18:57:42.873Z", "content": "Didn't take too long to find you on the leaderboard - top 5!! (at the time of writing) πͺ
", "rawMarkdown": "Didn't take too long to find you on the leaderboard - top 5!! (at the time of writing) πͺ", "votes": 1 } ] }, { "id": 1819219, "postDate": "2022-06-13T14:43:22.773Z", "content": "@datahobbit Thank you for sharing this awesome list!
", "rawMarkdown": "@datahobbit Thank you for sharing this awesome list!", "votes": 2 }, { "id": 1804321, "postDate": "2022-05-28T19:00:15.100Z", "rawMarkdown": "", "votes": 1, "isDeleted": true, "replies": [ { "id": 1804400, "postDate": "2022-05-28T22:19:09.013Z", "content": "You're welcome, that was a great competition, and hopefully, this will be too. It looks good so far!
", "rawMarkdown": "You're welcome, that was a great competition, and hopefully, this will be too. It looks good so far!" } ] }, { "id": 1884399, "postDate": "2022-08-04T12:06:53.323Z", "content": "thanks for sharing the list
", "rawMarkdown": "thanks for sharing the list", "votes": 1 } ], "comments": [ { "id": 1884359, "author_name": "Santosh_Kumar_Doodala", "author_url": "", "post_date": "2022-08-04T11:49:43.133000", "content": "ohh niceπ
", "votes": 1, "replies": [] }, { "id": 1818184, "author_name": "KritiDoneria", "author_url": "", "post_date": "2022-06-12T11:18:10.757000", "content": "@datahobbit , Thank you so much for considering Ethical considerations here :)
\nThanks
Sure no problem. This is becoming an increasingly important consideration with real-life models, especially those that can make a meaningful difference in the lives of individuals.
", "votes": 1, "replies": [] } ] }, { "id": 1807737, "author_name": "MIkhail Donskoy", "author_url": "", "post_date": "2022-06-01T09:27:32.917000", "content": "Hi ! Thanks for the source list.βοΈ
", "votes": 1, "replies": [ { "id": 1808202, "author_name": "MichaelP", "author_url": "", "post_date": "2022-06-01T15:53:44.780000", "content": "Happy to help.
", "votes": 0, "replies": [] } ] }, { "id": 1804737, "author_name": "Data_Wang", "author_url": "", "post_date": "2022-05-29T11:55:47.077000", "content": "πππππππππβοΈβοΈ
", "votes": 1, "replies": [] }, { "id": 1801999, "author_name": "NikhilMishra", "author_url": "", "post_date": "2022-05-26T10:34:23.773000", "content": "thanks for sharing this @datahobbit , looking to see you at the leaderboard
", "votes": 1, "replies": [ { "id": 1802472, "author_name": "MichaelP", "author_url": "", "post_date": "2022-05-26T18:57:42.873000", "content": "Didn't take too long to find you on the leaderboard - top 5!! (at the time of writing) πͺ
", "votes": 1, "replies": [] } ] }, { "id": 1819219, "author_name": "Dhamu", "author_url": "", "post_date": "2022-06-13T14:43:22.773000", "content": "@datahobbit Thank you for sharing this awesome list!
", "votes": 2, "replies": [] }, { "id": 1804321, "author_name": "", "author_url": "", "post_date": "2022-05-28T19:00:15.100000", "content": "", "votes": 1, "replies": [ { "id": 1804400, "author_name": "MichaelP", "author_url": "", "post_date": "2022-05-28T22:19:09.013000", "content": "You're welcome, that was a great competition, and hopefully, this will be too. It looks good so far!
", "votes": 0, "replies": [] } ] }, { "id": 1884399, "author_name": "Kirill Dzhunkovskii", "author_url": "", "post_date": "2022-08-04T12:06:53.323000", "content": "thanks for sharing the list
", "votes": 1, "replies": [] } ], "raw_markdown_by_id": { "1801527": "The Home Credit competition was the largest tabular data competition at the time on Kaggle and may yield some valuable insights. Like that competition, and default prediction in the real world, this is an unbalanced classification problem. As in most competitions, validation will be vitally important. Cross-validation may be particularly useful for unbalanced classification problems, but, it could be very time-consuming given the size of the data so sensible choices will need to be made about validation to ensure reasonable accuracy and speed.\n\nBelow are some further insights from the previous competition. Not all of the technical and modeling insights may translate to this competition but any domain insights and considerations about the time-series aspect of the competition may be very worthwhile to consider. \n\n[Reflections from a member of the winning team](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/58332)\n\n[First place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64821)\n\n[Summary of team member's contributions for 2nd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64722)\n\n[Some findings from the 2nd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64784)\n\n[3rd place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64596)\n\n[4th place solution and tips](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64487)\n\n[7th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64580)\n\n[8th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64474)\n\n[9th place solution and experience](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64536)\n\n[10th place solution](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/64598)\n\n[Ethical and Discrimination Considerations](https://www.kaggle.com/competitions/home-credit-default-risk/discussion/57028)\n\nThe features in the new competition do not fully spell out the meaning of the variables although the first letter of the feature name indicates the domain area where the feature belongs. That could be useful in trying to apply any domain insights.\n\nApologies if I missed any of the top solutions. Feel free to add any I missed or any other useful links or recollections from that competition.", "1884359": "ohh niceπ", "1818184": "@datahobbit , Thank you so much for considering Ethical considerations here :)\nThanks", "1807737": "Hi ! Thanks for the source list.βοΈ", "1804737": "πππππππππβοΈβοΈ", "1801999": "thanks for sharing this @datahobbit , looking to see you at the leaderboard ", "1819219": "@datahobbit Thank you for sharing this awesome list!", "1804321": "", "1884399": "thanks for sharing the list" } }