![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F761268%2F117d5e735e36677050b3a10e86d8994e%2FScreenshot%202024-01-20%20at%207.02.35AM.png?generation=1705714397991992&alt=media\")
Since, Other target dominate the targets, is it possible to split?
\n\n\nThe goal of this competition is to detect and classify seizures and other types of harmful brain activity in electroencephalography (EEG) data. Even experts find this to be a challenging task and often disagree about the correct labels.
\n
![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F761268%2F54d121e874b490594a7455bde0f637c7%2FScreenshot%202024-01-20%20at%206.42.06AM.png?generation=1705713703668675&alt=media\")
\n\nAll Targets are classified as 'Other' with respective to its distribution freq => is other target is weak label?
\n
TARGETS_GROUP2 = []\nTARGETS_GROUP1 = (((df.columns[-:])) - (TARGETS_GROUP2))\n==> modify both target groups to select which features are helpful to detect group1 vs group2.\nMaybe others can share with other models (wavnet, resnet, efficientnet) experiment results for binary classification?
\nNotebook - CatBoost Binary Classification: Other vs Rest copied from @cdeotte 's Notebook - CatBoost Starter - [LB 0.67]
", "messages": [ { "id": "2610145", "postDate": "01/19/2024 23:34:13", "content": "Since, Other target dominate the targets, is it possible to split?
\n\n\nThe goal of this competition is to detect and classify seizures and other types of harmful brain activity in electroencephalography (EEG) data. Even experts find this to be a challenging task and often disagree about the correct labels.
\n
\n\nAll Targets are classified as 'Other' with respective to its distribution freq => is other target is weak label?
\n
TARGETS_GROUP2 = []\nTARGETS_GROUP1 = (((df.columns[-:])) - (TARGETS_GROUP2))\n==> modify both target groups to select which features are helpful to detect group1 vs group2.\nMaybe others can share with other models (wavnet, resnet, efficientnet) experiment results for binary classification?
\nNotebook - CatBoost Binary Classification: Other vs Rest copied from @cdeotte 's Notebook - CatBoost Starter - [LB 0.67]
", "rawMarkdown": "**Since, Other target dominate the targets, is it possible to split?**\n> The goal of this competition is to **detect and classify seizures and other types of harmful brain activity** in electroencephalography (EEG) data. Even experts find this to be a challenging task and often disagree about the correct labels.\n\n\n\n# Predicted as \"Other\"\n\n\n> All Targets are classified as 'Other' with respective to its distribution freq => **is other target is weak label?**\n\n# How to approach the \"Other\" Target?\n\n```python\nTARGETS_GROUP2 = ['other_vote']\nTARGETS_GROUP1 = list(set(list(df.columns[-6:])) - set(TARGETS_GROUP2))\n==> modify both target groups to select which features are helpful to detect group1 vs group2.\n```\n\nMaybe others can share with other models (wavnet, resnet, efficientnet) experiment results for binary classification? \n\n[Notebook - CatBoost Binary Classification: Other vs Rest](https://www.kaggle.com/code/seshurajup/catboost-other-vs-all-brain-targets-classification/notebook) copied from @cdeotte 's [Notebook - CatBoost Starter - [LB 0.67]](https://www.kaggle.com/code/cdeotte/catboost-starter-lb-0-67?scriptVersionId=158817802)", "votes": null }, { "id": "2610200", "postDate": "01/20/2024 01:44:43", "content": "Another nice post. I like how you are constantly looking for answers. Keep going!
", "rawMarkdown": "Another nice post. I like how you are constantly looking for answers. Keep going!", "votes": null }, { "id": "2610205", "postDate": "01/20/2024 01:52:27", "content": "@yantxx i'm learning, just got motivated by @cdeotte. start seeing data in different perspective after learning power of mean experiment from Chris.
", "rawMarkdown": "yantxx i'm learning, just got motivated by @cdeotte. start seeing data in different perspective after learning *power of mean* experiment from Chris.", "votes": null } ], "comments": [ { "id": 2610200, "author_name": "yantxx", "author_url": "", "post_date": "01/20/2024 01:44:43", "content": "Another nice post. I like how you are constantly looking for answers. Keep going!
", "votes": null, "replies": [ { "id": 2610205, "author_name": "seshurajup", "author_url": "", "post_date": "01/20/2024 01:52:27", "content": "@yantxx i'm learning, just got motivated by @cdeotte. start seeing data in different perspective after learning power of mean experiment from Chris.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2610145": "**Since, Other target dominate the targets, is it possible to split?**\n> The goal of this competition is to **detect and classify seizures and other types of harmful brain activity** in electroencephalography (EEG) data. Even experts find this to be a challenging task and often disagree about the correct labels.\n\n\n\n# Predicted as \"Other\"\n\n\n> All Targets are classified as 'Other' with respective to its distribution freq => **is other target is weak label?**\n\n# How to approach the \"Other\" Target?\n\n```python\nTARGETS_GROUP2 = ['other_vote']\nTARGETS_GROUP1 = list(set(list(df.columns[-6:])) - set(TARGETS_GROUP2))\n==> modify both target groups to select which features are helpful to detect group1 vs group2.\n```\n\nMaybe others can share with other models (wavnet, resnet, efficientnet) experiment results for binary classification? \n\n[Notebook - CatBoost Binary Classification: Other vs Rest](https://www.kaggle.com/code/seshurajup/catboost-other-vs-all-brain-targets-classification/notebook) copied from @cdeotte 's [Notebook - CatBoost Starter - [LB 0.67]](https://www.kaggle.com/code/cdeotte/catboost-starter-lb-0-67?scriptVersionId=158817802)", "2610200": "Another nice post. I like how you are constantly looking for answers. Keep going!", "2610205": "yantxx i'm learning, just got motivated by @cdeotte. start seeing data in different perspective after learning *power of mean* experiment from Chris." }, "source": "meta" }