Hello everyone and welcome to this custom competition. The aim is to provide a longer, more diverse, tabular data competition, that could be both a learning place for newcomers and a benchmark to try and evaluate more advanced models.
\nThe competition is actually a 59-in-1 as the source for the datasets is the benchmark of the original paper: \"Why do tree-based models still outperform deep learning on tabular data?\". This should provide a diverse ensemble of tasks so that we can discuss general findings about tabular ML.
\nI'd like to thanks @carlmcbrideellis for beta-testing the competition and @leogri for discussions about the original paper. I would also like to thank the deepai image generator for the somewhat creepy illustration.
\nI hope you all enjoy this competition. Let's build, learn, and share new stuff about tabular ML!
", "messages": [ { "id": "2601955", "postDate": "01/14/2024 20:01:29", "content": "Hello everyone and welcome to this custom competition. The aim is to provide a longer, more diverse, tabular data competition, that could be both a learning place for newcomers and a benchmark to try and evaluate more advanced models.
\nThe competition is actually a 59-in-1 as the source for the datasets is the benchmark of the original paper: \"Why do tree-based models still outperform deep learning on tabular data?\". This should provide a diverse ensemble of tasks so that we can discuss general findings about tabular ML.
\nI'd like to thanks @carlmcbrideellis for beta-testing the competition and @leogri for discussions about the original paper. I would also like to thank the deepai image generator for the somewhat creepy illustration.
\nI hope you all enjoy this competition. Let's build, learn, and share new stuff about tabular ML!
", "rawMarkdown": "Hello everyone and welcome to this custom competition. The aim is to provide a longer, more diverse, tabular data competition, that could be both a learning place for newcomers and a benchmark to try and evaluate more advanced models. \n\nThe competition is actually a 59-in-1 as the source for the datasets is the benchmark of the original paper: \"Why do tree-based models still outperform deep learning on tabular data?\". This should provide a diverse ensemble of tasks so that we can discuss general findings about tabular ML. \n\nI'd like to thanks @carlmcbrideellis for beta-testing the competition and @leogri for discussions about the original paper. I would also like to thank the deepai image generator for the somewhat creepy illustration.\n\nI hope you all enjoy this competition. Let's build, learn, and share new stuff about tabular ML!", "votes": null }, { "id": "2601967", "postDate": "01/14/2024 20:06:01", "content": "@lucasmorin are we allowed to use tree methods or restrict to deep learning models only?
", "rawMarkdown": "lucasmorin are we allowed to use tree methods or restrict to deep learning models only?", "votes": null }, { "id": "2602000", "postDate": "01/14/2024 20:19:48", "content": "You are not restricted, quite the contrary… I'd like to see what everyone can come up with. If you can think of models outside of trees/NNs don't hesitate.
", "rawMarkdown": "You are not restricted, quite the contrary... I'd like to see what everyone can come up with. If you can think of models outside of trees/NNs don't hesitate.", "votes": null }, { "id": "2602264", "postDate": "01/15/2024 03:26:16", "content": "Thanks for the prompt rejoinder @lucasmorin
\nBest wishes to you for the competition!
thank you for posting this challenge Lucas. I'm keen on trying a few different approaches.
", "rawMarkdown": "thank you for posting this challenge Lucas. I'm keen on trying a few different approaches.", "votes": null } ], "comments": [ { "id": 2601967, "author_name": "ravi20076", "author_url": "", "post_date": "01/14/2024 20:06:01", "content": "@lucasmorin are we allowed to use tree methods or restrict to deep learning models only?
", "votes": null, "replies": [ { "id": 2602000, "author_name": "lucasmorin", "author_url": "", "post_date": "01/14/2024 20:19:48", "content": "You are not restricted, quite the contrary… I'd like to see what everyone can come up with. If you can think of models outside of trees/NNs don't hesitate.
", "votes": null, "replies": [ { "id": 2602264, "author_name": "ravi20076", "author_url": "", "post_date": "01/15/2024 03:26:16", "content": "Thanks for the prompt rejoinder @lucasmorin
\nBest wishes to you for the competition!
thank you for posting this challenge Lucas. I'm keen on trying a few different approaches.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2601955": "Hello everyone and welcome to this custom competition. The aim is to provide a longer, more diverse, tabular data competition, that could be both a learning place for newcomers and a benchmark to try and evaluate more advanced models. \n\nThe competition is actually a 59-in-1 as the source for the datasets is the benchmark of the original paper: \"Why do tree-based models still outperform deep learning on tabular data?\". This should provide a diverse ensemble of tasks so that we can discuss general findings about tabular ML. \n\nI'd like to thanks @carlmcbrideellis for beta-testing the competition and @leogri for discussions about the original paper. I would also like to thank the deepai image generator for the somewhat creepy illustration.\n\nI hope you all enjoy this competition. Let's build, learn, and share new stuff about tabular ML!", "2601967": "lucasmorin are we allowed to use tree methods or restrict to deep learning models only?", "2602000": "You are not restricted, quite the contrary... I'd like to see what everyone can come up with. If you can think of models outside of trees/NNs don't hesitate.", "2602264": "Thanks for the prompt rejoinder @lucasmorin \nBest wishes to you for the competition!", "2603350": "thank you for posting this challenge Lucas. I'm keen on trying a few different approaches." }, "source": "meta" }