{
  "id": 323170,
  "title": "Is ranker really better than Classifier?",
  "url": "/competitions/h-and-m-personalized-fashion-recommendations/discussion/323170",
  "author_name": "BearCater",
  "post_date": "2022-05-05T06:20:05.742000",
  "votes": 6,
  "comment_count": 9,
  "views": 0,
  "content": "<p>I don't know why, features? training data? label 1 or 0?  My ranker didn't work, can I try the classifier?</p>",
  "messages": [
    {
      "id": 1778130,
      "postDate": "2022-05-05T06:20:05.743Z",
      "content": "<p>I don't know why, features? training data? label 1 or 0?  My ranker didn't work, can I try the classifier?</p>",
      "rawMarkdown": "I don't know why, features? training data? label 1 or 0?  My ranker didn't work, can I try the classifier?",
      "votes": 6
    },
    {
      "id": 1778702,
      "postDate": "2022-05-05T15:58:02.293Z",
      "content": "<p>Classifier is not bad. I just have been using LGBMClassifier with logloss rather than LGBMRanker with lambdarank loss.</p>\n<p>(From another ECNUer 0w0)</p>",
      "rawMarkdown": "Classifier is not bad. I just have been using LGBMClassifier with logloss rather than LGBMRanker with lambdarank loss.\n\n(From another ECNUer 0w0)",
      "votes": 1,
      "replies": [
        {
          "id": 1778991,
          "postDate": "2022-05-06T01:39:13.760Z",
          "content": "<p>Thank you, I will try</p>",
          "rawMarkdown": "Thank you, I will try"
        },
        {
          "id": 1780355,
          "postDate": "2022-05-07T10:41:38.923Z",
          "content": "<p><a href=\"https://www.kaggle.com/sirius81\" target=\"_blank\">@sirius81</a> can you please share your CV logloss scores </p>",
          "rawMarkdown": "@sirius81 can you please share your CV logloss scores "
        },
        {
          "id": 1780379,
          "postDate": "2022-05-07T11:00:25.177Z",
          "content": "<p><a href=\"https://www.kaggle.com/tarique7\" target=\"_blank\">@tarique7</a> you ask MAP@12?  It's about 0.041x with the last week as the validation set. </p>",
          "rawMarkdown": "@tarique7 you ask MAP@12?  It's about 0.041x with the last week as the validation set. "
        },
        {
          "id": 1780452,
          "postDate": "2022-05-07T12:39:20.290Z",
          "content": "<p>I think I should have failed, but I also want to ask how you choose the sample of label 0, can you share the code after the competition, thank you very much!!!</p>",
          "rawMarkdown": "I think I should have failed, but I also want to ask how you choose the sample of label 0, can you share the code after the competition, thank you very much!!!"
        },
        {
          "id": 1780542,
          "postDate": "2022-05-07T14:35:52.777Z",
          "content": "<p><a href=\"https://www.kaggle.com/bearcater\" target=\"_blank\">@bearcater</a> See many detailed discussion in this <a href=\"https://www.kaggle.com/competitions/h-and-m-personalized-fashion-recommendations/discussion/307288\" target=\"_blank\">thread</a>. My strategy is the same: the generated candidates which the customer didn't purchase are labeled with 0 and those which the customer purchased are labeled with 1</p>",
          "rawMarkdown": "@bearcater See many detailed discussion in this [thread](https://www.kaggle.com/competitions/h-and-m-personalized-fashion-recommendations/discussion/307288). My strategy is the same: the generated candidates which the customer didn't purchase are labeled with 0 and those which the customer purchased are labeled with 1"
        }
      ]
    },
    {
      "id": 1778656,
      "postDate": "2022-05-05T15:01:03.143Z",
      "content": "<p>Your current LB is without any ranking?<br>\nThat's really good!</p>",
      "rawMarkdown": "Your current LB is without any ranking?\nThat's really good!",
      "votes": 1,
      "replies": [
        {
          "id": 1778996,
          "postDate": "2022-05-06T01:46:52.610Z",
          "content": "<p>far from you😭</p>",
          "rawMarkdown": "far from you😭"
        }
      ]
    },
    {
      "id": 1778974,
      "postDate": "2022-05-06T00:49:45.883Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1778702,
      "author_name": "sirius",
      "author_url": "",
      "post_date": "2022-05-05T15:58:02.293000",
      "content": "<p>Classifier is not bad. I just have been using LGBMClassifier with logloss rather than LGBMRanker with lambdarank loss.</p>\n<p>(From another ECNUer 0w0)</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1778991,
          "author_name": "BearCater",
          "author_url": "",
          "post_date": "2022-05-06T01:39:13.760000",
          "content": "<p>Thank you, I will try</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1780355,
          "author_name": "tarick.morty",
          "author_url": "",
          "post_date": "2022-05-07T10:41:38.923000",
          "content": "<p><a href=\"https://www.kaggle.com/sirius81\" target=\"_blank\">@sirius81</a> can you please share your CV logloss scores </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1780379,
          "author_name": "sirius",
          "author_url": "",
          "post_date": "2022-05-07T11:00:25.177000",
          "content": "<p><a href=\"https://www.kaggle.com/tarique7\" target=\"_blank\">@tarique7</a> you ask MAP@12?  It's about 0.041x with the last week as the validation set. </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1780452,
          "author_name": "BearCater",
          "author_url": "",
          "post_date": "2022-05-07T12:39:20.290000",
          "content": "<p>I think I should have failed, but I also want to ask how you choose the sample of label 0, can you share the code after the competition, thank you very much!!!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1780542,
          "author_name": "sirius",
          "author_url": "",
          "post_date": "2022-05-07T14:35:52.777000",
          "content": "<p><a href=\"https://www.kaggle.com/bearcater\" target=\"_blank\">@bearcater</a> See many detailed discussion in this <a href=\"https://www.kaggle.com/competitions/h-and-m-personalized-fashion-recommendations/discussion/307288\" target=\"_blank\">thread</a>. My strategy is the same: the generated candidates which the customer didn't purchase are labeled with 0 and those which the customer purchased are labeled with 1</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1778656,
      "author_name": "Clear n' Simple",
      "author_url": "",
      "post_date": "2022-05-05T15:01:03.143000",
      "content": "<p>Your current LB is without any ranking?<br>\nThat's really good!</p>",
      "votes": 1,
      "replies": [
        {
          "id": 1778996,
          "author_name": "BearCater",
          "author_url": "",
          "post_date": "2022-05-06T01:46:52.610000",
          "content": "<p>far from you😭</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1778974,
      "author_name": "",
      "author_url": "",
      "post_date": "2022-05-06T00:49:45.883000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1778130": "I don't know why, features? training data? label 1 or 0?  My ranker didn't work, can I try the classifier?",
    "1778702": "Classifier is not bad. I just have been using LGBMClassifier with logloss rather than LGBMRanker with lambdarank loss.\n\n(From another ECNUer 0w0)",
    "1778656": "Your current LB is without any ranking?\nThat's really good!",
    "1778974": ""
  }
}