{
  "id": 467626,
  "title": "LGBM LB[0.76] EEG and Spectrogram simple feature extraction",
  "url": "/competitions/hms-harmful-brain-activity-classification/discussion/467626",
  "author_name": "",
  "post_date": "2024-01-13T08:12:37.202835100Z",
  "votes": 9,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hello kagglers,<br>\nI recently published a starter notebook with public score of <a href=\"https://www.kaggle.com/code/arunsensei/lgbm-lb-0-76-spectrogram-and-eeg-describe\" target=\"_blank\">0.76</a>,<br>\neven tho the data given in this competition is large and complicated I wanted to see how simple feature extraction works on this data, so I extracted several features like min, max, std etcetera, I first made my experimental notebook with just using EEG feature extraction <a href=\"https://www.kaggle.com/code/arunsensei/lgbm-xgb-just-eeg-describe\" target=\"_blank\">0.96</a> and to my surprise it wasn't bad and it showed a bit potential while also handles nan/null values in the data, So I did the same feature extraction using  spectrograms and merged with eeg features and trained a LGBM model which got me to a good score in both CV and LB.<br>\nThanks for reading :')<br>\nleave an upvote on the notebooks and the post if you find any of this helpful,</p>",
  "messages": [
    {
      "id": "2599925",
      "postDate": "01/13/2024 08:12:37",
      "content": "<p>Hello kagglers,<br>\nI recently published a starter notebook with public score of <a href=\"https://www.kaggle.com/code/arunsensei/lgbm-lb-0-76-spectrogram-and-eeg-describe\" target=\"_blank\">0.76</a>,<br>\neven tho the data given in this competition is large and complicated I wanted to see how simple feature extraction works on this data, so I extracted several features like min, max, std etcetera, I first made my experimental notebook with just using EEG feature extraction <a href=\"https://www.kaggle.com/code/arunsensei/lgbm-xgb-just-eeg-describe\" target=\"_blank\">0.96</a> and to my surprise it wasn't bad and it showed a bit potential while also handles nan/null values in the data, So I did the same feature extraction using  spectrograms and merged with eeg features and trained a LGBM model which got me to a good score in both CV and LB.<br>\nThanks for reading :')<br>\nleave an upvote on the notebooks and the post if you find any of this helpful,</p>",
      "rawMarkdown": "Hello kagglers,\nI recently published a starter notebook with public score of [0.76](https://www.kaggle.com/code/arunsensei/lgbm-lb-0-76-spectrogram-and-eeg-describe),\neven tho the data given in this competition is large and complicated I wanted to see how simple feature extraction works on this data, so I extracted several features like min, max, std etcetera, I first made my experimental notebook with just using EEG feature extraction [0.96](https://www.kaggle.com/code/arunsensei/lgbm-xgb-just-eeg-describe) and to my surprise it wasn't bad and it showed a bit potential while also handles nan/null values in the data, So I did the same feature extraction using  spectrograms and merged with eeg features and trained a LGBM model which got me to a good score in both CV and LB.\nThanks for reading :')\nleave an upvote on the notebooks and the post if you find any of this helpful,",
      "votes": null
    },
    {
      "id": "2600100",
      "postDate": "01/13/2024 11:18:23",
      "content": "<p>Great work <a href=\"https://www.kaggle.com/arunsensei\" target=\"_blank\">@arunsensei</a> Thanks for sharing!</p>",
      "rawMarkdown": "Great work @arunsensei Thanks for sharing!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2600100,
      "author_name": "cdeotte",
      "author_url": "",
      "post_date": "01/13/2024 11:18:23",
      "content": "<p>Great work <a href=\"https://www.kaggle.com/arunsensei\" target=\"_blank\">@arunsensei</a> Thanks for sharing!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2599925": "Hello kagglers,\nI recently published a starter notebook with public score of [0.76](https://www.kaggle.com/code/arunsensei/lgbm-lb-0-76-spectrogram-and-eeg-describe),\neven tho the data given in this competition is large and complicated I wanted to see how simple feature extraction works on this data, so I extracted several features like min, max, std etcetera, I first made my experimental notebook with just using EEG feature extraction [0.96](https://www.kaggle.com/code/arunsensei/lgbm-xgb-just-eeg-describe) and to my surprise it wasn't bad and it showed a bit potential while also handles nan/null values in the data, So I did the same feature extraction using  spectrograms and merged with eeg features and trained a LGBM model which got me to a good score in both CV and LB.\nThanks for reading :')\nleave an upvote on the notebooks and the post if you find any of this helpful,",
    "2600100": "Great work @arunsensei Thanks for sharing!"
  },
  "source": "meta"
}