{
  "id": 69580,
  "title": "Skewed data ",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/69580",
  "author_name": "sumit",
  "post_date": "2018-10-25T04:36:46.115000",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>HI</p>\n\n<p>While analyzing the data I come to know that only 1.2% (approx) of the total customers contribute the revenue. </p>\n\n<p>This means that our data is unbalanced and skewed.</p>\n\n<p>So , will it be possible to use LightGBM directly without balancing the data or do I need to balance data first and then use LightGBM or any another algorithm  </p>",
  "messages": [
    {
      "id": 409926,
      "postDate": "2018-10-25T04:36:46.117Z",
      "content": "<p>HI</p>\n\n<p>While analyzing the data I come to know that only 1.2% (approx) of the total customers contribute the revenue. </p>\n\n<p>This means that our data is unbalanced and skewed.</p>\n\n<p>So , will it be possible to use LightGBM directly without balancing the data or do I need to balance data first and then use LightGBM or any another algorithm  </p>",
      "rawMarkdown": "HI\n\nWhile analyzing the data I come to know that only 1.2% (approx) of the total customers contribute the revenue. \n\nThis means that our data is unbalanced and skewed.\n\nSo , will it be possible to use LightGBM directly without balancing the data or do I need to balance data first and then use LightGBM or any another algorithm  ",
      "votes": 1
    },
    {
      "id": 410733,
      "postDate": "2018-10-26T15:10:11.850Z",
      "content": "<p>Is there actually an issue? I think skewed data is an issue in classification problems as it makes naive accuracy metrics fail. But this is not a concern for regression with a RMSE error as we have here, especially with a model with low bias, like LightGBM.</p>",
      "rawMarkdown": "Is there actually an issue? I think skewed data is an issue in classification problems as it makes naive accuracy metrics fail. But this is not a concern for regression with a RMSE error as we have here, especially with a model with low bias, like LightGBM."
    },
    {
      "id": 411178,
      "postDate": "2018-10-27T14:58:29.143Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 410733,
      "author_name": "Burrito Dan",
      "author_url": "",
      "post_date": "2018-10-26T15:10:11.850000",
      "content": "<p>Is there actually an issue? I think skewed data is an issue in classification problems as it makes naive accuracy metrics fail. But this is not a concern for regression with a RMSE error as we have here, especially with a model with low bias, like LightGBM.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 411178,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-10-27T14:58:29.143000",
      "content": "",
      "votes": 1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "409926": "HI\n\nWhile analyzing the data I come to know that only 1.2% (approx) of the total customers contribute the revenue. \n\nThis means that our data is unbalanced and skewed.\n\nSo , will it be possible to use LightGBM directly without balancing the data or do I need to balance data first and then use LightGBM or any another algorithm  ",
    "410733": "Is there actually an issue? I think skewed data is an issue in classification problems as it makes naive accuracy metrics fail. But this is not a concern for regression with a RMSE error as we have here, especially with a model with low bias, like LightGBM.",
    "411178": ""
  }
}