What’s the deal with this dataset? It’s so skewed, it’s like it’s been through a funhouse mirror! You think basic cleaning and stacking models are gonna fix this? That’s like trying to fix a flat tire with duct tape!
\nAnd 34 severe cases with class 3? Come on! That’s like ordering a pizza and getting just one slice! We need to get creative here—let’s figure out how to actually make sense of this thing, because right now, it’s a hot mess!
", "messages": [ { "id": "3016007", "postDate": "10/13/2024 08:42:59", "content": "What’s the deal with this dataset? It’s so skewed, it’s like it’s been through a funhouse mirror! You think basic cleaning and stacking models are gonna fix this? That’s like trying to fix a flat tire with duct tape!
\nAnd 34 severe cases with class 3? Come on! That’s like ordering a pizza and getting just one slice! We need to get creative here—let’s figure out how to actually make sense of this thing, because right now, it’s a hot mess!
", "rawMarkdown": "What’s the deal with this dataset? It’s so skewed, it’s like it’s been through a funhouse mirror! You think basic cleaning and stacking models are gonna fix this? That’s like trying to fix a flat tire with duct tape!\n\nAnd 34 severe cases with class 3? Come on! That’s like ordering a pizza and getting just one slice! We need to get creative here—let’s figure out how to actually make sense of this thing, because right now, it’s a hot mess!", "votes": null }, { "id": "3016102", "postDate": "10/13/2024 10:55:24", "content": "https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538235
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535525
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535900
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535354
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536407
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537514
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535321
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535440
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538243
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535226
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537169
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535388
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536694
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535956
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536321
I can't really provide much information or help regarding problems with the datasets provided but here are some discussions may/may not related to concerns towards the datasets
", "rawMarkdown": "[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538235](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535525](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535900](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535354](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536407](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537514](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535321](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535440](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538243](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535226](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537169](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535388](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536694](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535956](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536321](url)\n\nI can't really provide much information or help regarding problems with the datasets provided but here are some discussions may/may not related to concerns towards the datasets", "votes": null }, { "id": "3016332", "postDate": "10/13/2024 16:07:43", "content": "what are you looking for if you're now in the top rank. Congratulations!
\nI guess feature interaction is key. None of the feature is decisive. lgbm etc can capture most of the interaction very nicely but I guess they're still missing something important.
\nWhat brings you to the current rank? Ensembling, hyperparameter tuning or feature engineering?
Thanks! I’m looking to explore more creative approaches to handling this data, as it’s as real as it gets. It feels like a great opportunity to learn something practical.
", "rawMarkdown": "Thanks! I’m looking to explore more creative approaches to handling this data, as it’s as real as it gets. It feels like a great opportunity to learn something practical.", "votes": null }, { "id": "3016522", "postDate": "10/13/2024 22:17:18", "content": "You could try using a GAN to generate new data.
", "rawMarkdown": "You could try using a GAN to generate new data.", "votes": null }, { "id": "3016526", "postDate": "10/13/2024 22:39:46", "content": "Wow. This is the 1st time I hear about GAN, amazing and fun. Thank you for sharing.
\nhttps://www.kaggle.com/code/jesucristo/gan-introduction
", "rawMarkdown": "Wow. This is the 1st time I hear about GAN, amazing and fun. Thank you for sharing.\n\nhttps://www.kaggle.com/code/jesucristo/gan-introduction", "votes": null }, { "id": "3016528", "postDate": "10/13/2024 22:53:23", "content": "I am actually working on GAN for next few submissions.
", "rawMarkdown": "I am actually working on GAN for next few submissions.", "votes": null }, { "id": "3016530", "postDate": "10/13/2024 22:54:55", "content": "Thanks to Ian Goodfellow. GAN is indeed cool and fun.
", "rawMarkdown": "Thanks to Ian Goodfellow. GAN is indeed cool and fun.", "votes": null }, { "id": "3017674", "postDate": "10/15/2024 05:34:29", "content": "For code examples of GAN, you can check out Kaggles previous Dog GAN Comp.
", "rawMarkdown": "For code examples of GAN, you can check out Kaggles previous Dog GAN Comp.", "votes": null }, { "id": "3017753", "postDate": "10/15/2024 07:12:51", "content": "Thanks. I actually have already implemented it. Not very useful as of now… experimenting with hyperparameters.
\nI was looking into this competition seem very interested https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting
\nLet me know if you want to team up on same. We can have a discussion if you are interested too.
", "rawMarkdown": "Thanks. I actually have already implemented it. Not very useful as of now... experimenting with hyperparameters.\n\nI was looking into this competition seem very interested https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting\n\nLet me know if you want to team up on same. We can have a discussion if you are interested too.", "votes": null }, { "id": "3018091", "postDate": "10/15/2024 14:05:38", "content": "diffusion models crying now no one looking at them 😅
", "rawMarkdown": "diffusion models crying now no one looking at them 😅", "votes": null }, { "id": "3018175", "postDate": "10/15/2024 15:07:06", "content": "okay. Just because you mention it, I hereby pledge my next 3 submissions with diffusion models.
", "rawMarkdown": "okay. Just because you mention it, I hereby pledge my next 3 submissions with diffusion models.", "votes": null }, { "id": "3018224", "postDate": "10/15/2024 16:12:04", "content": "i tried them but nothing new based on this paper https://arxiv.org/pdf/2309.09968
", "rawMarkdown": "i tried them but nothing new based on this paper https://arxiv.org/pdf/2309.09968", "votes": null } ], "comments": [ { "id": 3016102, "author_name": "cchangyyy", "author_url": "", "post_date": "10/13/2024 10:55:24", "content": "https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538235
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535525
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535900
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535354
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536407
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537514
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535321
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535440
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538243
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535226
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537169
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535388
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536694
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535956
\nhttps://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536321
I can't really provide much information or help regarding problems with the datasets provided but here are some discussions may/may not related to concerns towards the datasets
", "votes": null, "replies": [] }, { "id": 3016332, "author_name": "tomyuen", "author_url": "", "post_date": "10/13/2024 16:07:43", "content": "what are you looking for if you're now in the top rank. Congratulations!
\nI guess feature interaction is key. None of the feature is decisive. lgbm etc can capture most of the interaction very nicely but I guess they're still missing something important.
\nWhat brings you to the current rank? Ensembling, hyperparameter tuning or feature engineering?
Thanks! I’m looking to explore more creative approaches to handling this data, as it’s as real as it gets. It feels like a great opportunity to learn something practical.
", "votes": null, "replies": [ { "id": 3016522, "author_name": "peterhopkinson", "author_url": "", "post_date": "10/13/2024 22:17:18", "content": "You could try using a GAN to generate new data.
", "votes": null, "replies": [ { "id": 3016526, "author_name": "tomyuen", "author_url": "", "post_date": "10/13/2024 22:39:46", "content": "Wow. This is the 1st time I hear about GAN, amazing and fun. Thank you for sharing.
\nhttps://www.kaggle.com/code/jesucristo/gan-introduction
", "votes": null, "replies": [ { "id": 3016530, "author_name": "chanpreetsingh07", "author_url": "", "post_date": "10/13/2024 22:54:55", "content": "Thanks to Ian Goodfellow. GAN is indeed cool and fun.
", "votes": null, "replies": [] } ] }, { "id": 3016528, "author_name": "chanpreetsingh07", "author_url": "", "post_date": "10/13/2024 22:53:23", "content": "I am actually working on GAN for next few submissions.
", "votes": null, "replies": [] } ] }, { "id": 3017674, "author_name": "cdeotte", "author_url": "", "post_date": "10/15/2024 05:34:29", "content": "For code examples of GAN, you can check out Kaggles previous Dog GAN Comp.
", "votes": null, "replies": [ { "id": 3017753, "author_name": "chanpreetsingh07", "author_url": "", "post_date": "10/15/2024 07:12:51", "content": "Thanks. I actually have already implemented it. Not very useful as of now… experimenting with hyperparameters.
\nI was looking into this competition seem very interested https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting
\nLet me know if you want to team up on same. We can have a discussion if you are interested too.
", "votes": null, "replies": [ { "id": 3018091, "author_name": "riadalmadani", "author_url": "", "post_date": "10/15/2024 14:05:38", "content": "diffusion models crying now no one looking at them 😅
", "votes": null, "replies": [ { "id": 3018175, "author_name": "chanpreetsingh07", "author_url": "", "post_date": "10/15/2024 15:07:06", "content": "okay. Just because you mention it, I hereby pledge my next 3 submissions with diffusion models.
", "votes": null, "replies": [ { "id": 3018224, "author_name": "riadalmadani", "author_url": "", "post_date": "10/15/2024 16:12:04", "content": "i tried them but nothing new based on this paper https://arxiv.org/pdf/2309.09968
", "votes": null, "replies": [] } ] } ] } ] } ] } ] } ] } ], "raw_markdown_by_id": { "3016007": "What’s the deal with this dataset? It’s so skewed, it’s like it’s been through a funhouse mirror! You think basic cleaning and stacking models are gonna fix this? That’s like trying to fix a flat tire with duct tape!\n\nAnd 34 severe cases with class 3? Come on! That’s like ordering a pizza and getting just one slice! We need to get creative here—let’s figure out how to actually make sense of this thing, because right now, it’s a hot mess!", "3016102": "[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538235](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535525](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535900](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535354](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536407](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537514](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535321](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535440](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/538243](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535226](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/537169](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535388](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536694](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/535956](url)\n[https://www.kaggle.com/competitions/child-mind-institute-problematic-internet-use/discussion/536321](url)\n\nI can't really provide much information or help regarding problems with the datasets provided but here are some discussions may/may not related to concerns towards the datasets", "3016332": "what are you looking for if you're now in the top rank. Congratulations!\nI guess feature interaction is key. None of the feature is decisive. lgbm etc can capture most of the interaction very nicely but I guess they're still missing something important. \nWhat brings you to the current rank? Ensembling, hyperparameter tuning or feature engineering?", "3016399": "Thanks! I’m looking to explore more creative approaches to handling this data, as it’s as real as it gets. It feels like a great opportunity to learn something practical.", "3016522": "You could try using a GAN to generate new data.", "3016526": "Wow. This is the 1st time I hear about GAN, amazing and fun. Thank you for sharing.\n\nhttps://www.kaggle.com/code/jesucristo/gan-introduction", "3016528": "I am actually working on GAN for next few submissions.", "3016530": "Thanks to Ian Goodfellow. GAN is indeed cool and fun.", "3017674": "For code examples of GAN, you can check out Kaggles previous Dog GAN Comp.", "3017753": "Thanks. I actually have already implemented it. Not very useful as of now... experimenting with hyperparameters.\n\nI was looking into this competition seem very interested https://www.kaggle.com/competitions/jane-street-real-time-market-data-forecasting\n\nLet me know if you want to team up on same. We can have a discussion if you are interested too.", "3018091": "diffusion models crying now no one looking at them 😅", "3018175": "okay. Just because you mention it, I hereby pledge my next 3 submissions with diffusion models.", "3018224": "i tried them but nothing new based on this paper https://arxiv.org/pdf/2309.09968" }, "source": "meta" }