{
  "id": 307472,
  "title": "🎯 What's your CV Strategy? ",
  "url": "/competitions/happy-whale-and-dolphin/discussion/307472",
  "author_name": "",
  "post_date": "2022-02-14T10:24:54.259316200Z",
  "votes": 21,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Hello everyone! </p>\n<p>I'm still working on setting up my CV effectively. What's your strategy for this competition?</p>\n<p>Also, what are your CV/LB scores, if you're okay to share them :)</p>\n<p>Thanks in advance!</p>",
  "messages": [
    {
      "id": "1689561",
      "postDate": "02/14/2022 10:24:54",
      "content": "<p>Hello everyone! </p>\n<p>I'm still working on setting up my CV effectively. What's your strategy for this competition?</p>\n<p>Also, what are your CV/LB scores, if you're okay to share them :)</p>\n<p>Thanks in advance!</p>",
      "rawMarkdown": "Hello everyone! \n\nI'm still working on setting up my CV effectively. What's your strategy for this competition?\n\nAlso, what are your CV/LB scores, if you're okay to share them :)\n\nThanks in advance!",
      "votes": null
    },
    {
      "id": "1690171",
      "postDate": "02/14/2022 18:52:42",
      "content": "<p>I am using  starityfiedKfolds on individual id, where k=5<br>\nwill update  later my CV and LB score</p>",
      "rawMarkdown": "I am using  starityfiedKfolds on individual id, where k=5\nwill update  later my CV and LB score",
      "votes": null
    },
    {
      "id": "1690700",
      "postDate": "02/15/2022 04:13:13",
      "content": "<p>I am yet to fully implemented my first model 😬 but for folds I used multi label stratifiedkfold with species and individual_ids. However want to see the distribution visually, bit unsure if I could do better. Also thinking if I should remove the ids with single instance</p>",
      "rawMarkdown": "I am yet to fully implemented my first model 😬 but for folds I used multi label stratifiedkfold with species and individual_ids. However want to see the distribution visually, bit unsure if I could do better. Also thinking if I should remove the ids with single instance",
      "votes": null
    },
    {
      "id": "1695879",
      "postDate": "02/18/2022 12:28:26",
      "content": "<p>This is the most important question!<br>\n59.4% of the dataset contains only one image per id.<br>\nWould you like to share your strategy?😃</p>",
      "rawMarkdown": "This is the most important question!\n59.4% of the dataset contains only one image per id.\nWould you like to share your strategy?😃",
      "votes": null
    },
    {
      "id": "1695885",
      "postDate": "02/18/2022 12:34:56",
      "content": "<p><a href=\"https://www.kaggle.com/rajborgohain\" target=\"_blank\">@rajborgohain</a> I'm still working on establishing a baseline and asked to learn from others 😄</p>",
      "rawMarkdown": "rajborgohain I'm still working on establishing a baseline and asked to learn from others 😄",
      "votes": null
    },
    {
      "id": "1696146",
      "postDate": "02/18/2022 16:10:02",
      "content": "<p><code>59.4% of the dataset contains only one image per id.</code><br>\nhow? 9k+ images out of 50k images have 1 single id, in train data</p>",
      "rawMarkdown": "`59.4% of the dataset contains only one image per id.`\nhow? 9k+ images out of 50k images have 1 single id, in train data",
      "votes": null
    },
    {
      "id": "1696165",
      "postDate": "02/18/2022 16:24:30",
      "content": "<p>Yes, it's 9258, and 5573 of them are whales, and 3685 are dolphin</p>",
      "rawMarkdown": "Yes, it's 9258, and 5573 of them are whales, and 3685 are dolphin",
      "votes": null
    },
    {
      "id": "1749111",
      "postDate": "04/08/2022 08:19:27",
      "content": "<p>👀Besides startifiedkfold of individual ids, is there any better CV strategy?(Is this OK to ask?)</p>",
      "rawMarkdown": "👀Besides startifiedkfold of individual ids, is there any better CV strategy?(Is this OK to ask?)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1690171,
      "author_name": "mrinath",
      "author_url": "",
      "post_date": "02/14/2022 18:52:42",
      "content": "<p>I am using  starityfiedKfolds on individual id, where k=5<br>\nwill update  later my CV and LB score</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1690700,
      "author_name": "bsridatta",
      "author_url": "",
      "post_date": "02/15/2022 04:13:13",
      "content": "<p>I am yet to fully implemented my first model 😬 but for folds I used multi label stratifiedkfold with species and individual_ids. However want to see the distribution visually, bit unsure if I could do better. Also thinking if I should remove the ids with single instance</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1695879,
      "author_name": "rajborgohain",
      "author_url": "",
      "post_date": "02/18/2022 12:28:26",
      "content": "<p>This is the most important question!<br>\n59.4% of the dataset contains only one image per id.<br>\nWould you like to share your strategy?😃</p>",
      "votes": null,
      "replies": [
        {
          "id": 1695885,
          "author_name": "init27",
          "author_url": "",
          "post_date": "02/18/2022 12:34:56",
          "content": "<p><a href=\"https://www.kaggle.com/rajborgohain\" target=\"_blank\">@rajborgohain</a> I'm still working on establishing a baseline and asked to learn from others 😄</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1696146,
          "author_name": "mrinath",
          "author_url": "",
          "post_date": "02/18/2022 16:10:02",
          "content": "<p><code>59.4% of the dataset contains only one image per id.</code><br>\nhow? 9k+ images out of 50k images have 1 single id, in train data</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1696165,
          "author_name": "rajborgohain",
          "author_url": "",
          "post_date": "02/18/2022 16:24:30",
          "content": "<p>Yes, it's 9258, and 5573 of them are whales, and 3685 are dolphin</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1749111,
      "author_name": "jimmysmith1009",
      "author_url": "",
      "post_date": "04/08/2022 08:19:27",
      "content": "<p>👀Besides startifiedkfold of individual ids, is there any better CV strategy?(Is this OK to ask?)</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1689561": "Hello everyone! \n\nI'm still working on setting up my CV effectively. What's your strategy for this competition?\n\nAlso, what are your CV/LB scores, if you're okay to share them :)\n\nThanks in advance!",
    "1690171": "I am using  starityfiedKfolds on individual id, where k=5\nwill update  later my CV and LB score",
    "1690700": "I am yet to fully implemented my first model 😬 but for folds I used multi label stratifiedkfold with species and individual_ids. However want to see the distribution visually, bit unsure if I could do better. Also thinking if I should remove the ids with single instance",
    "1695879": "This is the most important question!\n59.4% of the dataset contains only one image per id.\nWould you like to share your strategy?😃",
    "1695885": "rajborgohain I'm still working on establishing a baseline and asked to learn from others 😄",
    "1696146": "`59.4% of the dataset contains only one image per id.`\nhow? 9k+ images out of 50k images have 1 single id, in train data",
    "1696165": "Yes, it's 9258, and 5573 of them are whales, and 3685 are dolphin",
    "1749111": "👀Besides startifiedkfold of individual ids, is there any better CV strategy?(Is this OK to ask?)"
  },
  "source": "meta"
}