{
  "id": 261659,
  "title": "what about normalization?",
  "url": "/competitions/seti-breakthrough-listen/discussion/261659",
  "author_name": "Aristarkh_BFG",
  "post_date": "2021-08-04T19:29:55.141000",
  "votes": 3,
  "comment_count": 5,
  "views": 0,
  "content": "<p>which data normalization strategy works best for you?</p>",
  "messages": [
    {
      "id": 1454537,
      "postDate": "2021-08-06T08:29:25.113Z",
      "content": "<p>New data is normalized.  If you use old data you should normalize it to 0 mean and variance 1 as it wasn't properly normalized, see <a href=\"https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194\" target=\"_blank\">https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194</a></p>",
      "rawMarkdown": "New data is normalized.  If you use old data you should normalize it to 0 mean and variance 1 as it wasn't properly normalized, see https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194",
      "votes": 6,
      "replies": [
        {
          "id": 1454582,
          "postDate": "2021-08-06T08:53:52.613Z",
          "content": "<p>thank you for the answer)</p>",
          "rawMarkdown": "thank you for the answer)",
          "votes": 1
        }
      ]
    },
    {
      "id": 1449012,
      "postDate": "2021-08-04T19:29:55.143Z",
      "content": "<p>which data normalization strategy works best for you?</p>",
      "rawMarkdown": "which data normalization strategy works best for you?",
      "votes": 3
    },
    {
      "id": 1452122,
      "postDate": "2021-08-05T14:22:51.017Z",
      "content": "<p>applying no normalization works fine for me… but you can do scaling (image/255)</p>",
      "rawMarkdown": "applying no normalization works fine for me... but you can do scaling (image/255)",
      "votes": 1,
      "replies": [
        {
          "id": 1452210,
          "postDate": "2021-08-05T14:45:24.557Z",
          "content": "<p>the data is already normalized to zero mean and a standard deviation of one which is usually a good input for neural nets. </p>",
          "rawMarkdown": "the data is already normalized to zero mean and a standard deviation of one which is usually a good input for neural nets. ",
          "votes": 18
        },
        {
          "id": 1453338,
          "postDate": "2021-08-05T20:35:29.177Z",
          "content": "<p>Yeah you are right I checked it! scaling does not make sense. I am using it without normalization and scaling. </p>",
          "rawMarkdown": "Yeah you are right I checked it! scaling does not make sense. I am using it without normalization and scaling. "
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1454537,
      "author_name": "CPMP",
      "author_url": "",
      "post_date": "2021-08-06T08:29:25.113000",
      "content": "<p>New data is normalized.  If you use old data you should normalize it to 0 mean and variance 1 as it wasn't properly normalized, see <a href=\"https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194\" target=\"_blank\">https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194</a></p>",
      "votes": 6,
      "replies": [
        {
          "id": 1454582,
          "author_name": "Aristarkh_BFG",
          "author_url": "",
          "post_date": "2021-08-06T08:53:52.613000",
          "content": "<p>thank you for the answer)</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 1452122,
      "author_name": "Parth Dhameliya",
      "author_url": "",
      "post_date": "2021-08-05T14:22:51.017000",
      "content": "<p>applying no normalization works fine for me… but you can do scaling (image/255)</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1452210,
          "author_name": "Pascal Pfeiffer",
          "author_url": "",
          "post_date": "2021-08-05T14:45:24.557000",
          "content": "<p>the data is already normalized to zero mean and a standard deviation of one which is usually a good input for neural nets. </p>",
          "votes": 18,
          "replies": []
        },
        {
          "id": 1453338,
          "author_name": "Parth Dhameliya",
          "author_url": "",
          "post_date": "2021-08-05T20:35:29.177000",
          "content": "<p>Yeah you are right I checked it! scaling does not make sense. I am using it without normalization and scaling. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1454537": "New data is normalized.  If you use old data you should normalize it to 0 mean and variance 1 as it wasn't properly normalized, see https://www.kaggle.com/c/seti-breakthrough-listen/discussion/248194",
    "1449012": "which data normalization strategy works best for you?",
    "1452122": "applying no normalization works fine for me... but you can do scaling (image/255)"
  }
}