{
  "id": 333575,
  "title": "\"the probability of a future payment default (target = 1)\" means?",
  "url": "/competitions/amex-default-prediction/discussion/333575",
  "author_name": "",
  "post_date": "2022-06-27T09:11:41.114763900Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>While I am reading the description,<br>\nI found this statement</p>\n<p><strong>\"the probability of a future payment default (target = 1)\"</strong></p>\n<p>It seems like the target we try to predict is not clear because of the statement.</p>\n<p>Do we submit the prediction 0 or 1?<br>\nex)<br>\ncustomer id1: 0<br>\ncustomer id2: 1<br>\ncustomer id3: 0<br>\ncustomer id4: 0<br>\ncustomer id5: 1</p>\n<p>or some probability ranged from 0 to 1 that could possibly occur default(target=1)?<br>\nlike 0.79 means 79% of chance to occur default..<br>\n0.1 means 10% of chance it would occur default.., etc.<br>\nex)<br>\ncustomer id1: 0.03<br>\ncustomer id2: 0.95<br>\ncustomer id3: 0.12<br>\ncustomer id4: 0.19<br>\ncustomer id5: 0.86</p>",
  "messages": [
    {
      "id": "1834876",
      "postDate": "06/27/2022 09:11:41",
      "content": "<p>While I am reading the description,<br>\nI found this statement</p>\n<p><strong>\"the probability of a future payment default (target = 1)\"</strong></p>\n<p>It seems like the target we try to predict is not clear because of the statement.</p>\n<p>Do we submit the prediction 0 or 1?<br>\nex)<br>\ncustomer id1: 0<br>\ncustomer id2: 1<br>\ncustomer id3: 0<br>\ncustomer id4: 0<br>\ncustomer id5: 1</p>\n<p>or some probability ranged from 0 to 1 that could possibly occur default(target=1)?<br>\nlike 0.79 means 79% of chance to occur default..<br>\n0.1 means 10% of chance it would occur default.., etc.<br>\nex)<br>\ncustomer id1: 0.03<br>\ncustomer id2: 0.95<br>\ncustomer id3: 0.12<br>\ncustomer id4: 0.19<br>\ncustomer id5: 0.86</p>",
      "rawMarkdown": "While I am reading the description,\nI found this statement\n\n**\"the probability of a future payment default (target = 1)\"**\n\nIt seems like the target we try to predict is not clear because of the statement.\n\nDo we submit the prediction 0 or 1?\nex)\ncustomer id1: 0\ncustomer id2: 1\ncustomer id3: 0\ncustomer id4: 0\ncustomer id5: 1\n\nor some probability ranged from 0 to 1 that could possibly occur default(target=1)?\nlike 0.79 means 79% of chance to occur default..\n0.1 means 10% of chance it would occur default.., etc.\nex)\ncustomer id1: 0.03\ncustomer id2: 0.95\ncustomer id3: 0.12\ncustomer id4: 0.19\ncustomer id5: 0.86",
      "votes": null
    },
    {
      "id": "1834883",
      "postDate": "06/27/2022 09:23:58",
      "content": "<p>In fact, both of them are ok. You can even submit numbers that are not between 0 and 1.  Because, the more general evaluation metric AUC, only focuses on sorting</p>",
      "rawMarkdown": "In fact, both of them are ok. You can even submit numbers that are not between 0 and 1.  Because, the more general evaluation metric AUC, only focuses on sorting",
      "votes": null
    },
    {
      "id": "1835045",
      "postDate": "06/27/2022 12:06:55",
      "content": "<p>It is not necessary to fit the prediction results from 0 to 1, but I think it is better to rank them because of D (the default rate captured at 4%).<br>\nI wrote a little about the metric. It might be able to help.<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/333338\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/333338</a></p>",
      "rawMarkdown": "It is not necessary to fit the prediction results from 0 to 1, but I think it is better to rank them because of D (the default rate captured at 4%).\nI wrote a little about the metric. It might be able to help.\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/333338",
      "votes": null
    },
    {
      "id": "1838031",
      "postDate": "06/30/2022 06:33:26",
      "content": "<p>Ok … How do they then calculate the public score? Do they sort based on the score we give then match top n and count how many of them are actually ended up with fault?</p>",
      "rawMarkdown": "Ok ... How do they then calculate the public score? Do they sort based on the score we give then match top n and count how many of them are actually ended up with fault?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1834883,
      "author_name": "mahluo",
      "author_url": "",
      "post_date": "06/27/2022 09:23:58",
      "content": "<p>In fact, both of them are ok. You can even submit numbers that are not between 0 and 1.  Because, the more general evaluation metric AUC, only focuses on sorting</p>",
      "votes": null,
      "replies": [
        {
          "id": 1838031,
          "author_name": "gilgarad",
          "author_url": "",
          "post_date": "06/30/2022 06:33:26",
          "content": "<p>Ok … How do they then calculate the public score? Do they sort based on the score we give then match top n and count how many of them are actually ended up with fault?</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1835045,
      "author_name": "hanaori",
      "author_url": "",
      "post_date": "06/27/2022 12:06:55",
      "content": "<p>It is not necessary to fit the prediction results from 0 to 1, but I think it is better to rank them because of D (the default rate captured at 4%).<br>\nI wrote a little about the metric. It might be able to help.<br>\n<a href=\"https://www.kaggle.com/competitions/amex-default-prediction/discussion/333338\" target=\"_blank\">https://www.kaggle.com/competitions/amex-default-prediction/discussion/333338</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1834876": "While I am reading the description,\nI found this statement\n\n**\"the probability of a future payment default (target = 1)\"**\n\nIt seems like the target we try to predict is not clear because of the statement.\n\nDo we submit the prediction 0 or 1?\nex)\ncustomer id1: 0\ncustomer id2: 1\ncustomer id3: 0\ncustomer id4: 0\ncustomer id5: 1\n\nor some probability ranged from 0 to 1 that could possibly occur default(target=1)?\nlike 0.79 means 79% of chance to occur default..\n0.1 means 10% of chance it would occur default.., etc.\nex)\ncustomer id1: 0.03\ncustomer id2: 0.95\ncustomer id3: 0.12\ncustomer id4: 0.19\ncustomer id5: 0.86",
    "1834883": "In fact, both of them are ok. You can even submit numbers that are not between 0 and 1.  Because, the more general evaluation metric AUC, only focuses on sorting",
    "1835045": "It is not necessary to fit the prediction results from 0 to 1, but I think it is better to rank them because of D (the default rate captured at 4%).\nI wrote a little about the metric. It might be able to help.\nhttps://www.kaggle.com/competitions/amex-default-prediction/discussion/333338",
    "1838031": "Ok ... How do they then calculate the public score? Do they sort based on the score we give then match top n and count how many of them are actually ended up with fault?"
  },
  "source": "meta"
}