{
  "id": 307202,
  "title": "Users that bought more than 12 items during the test period",
  "url": "/competitions/h-and-m-personalized-fashion-recommendations/discussion/307202",
  "author_name": "Victor Gonzalez",
  "post_date": "2022-02-13T07:50:17.587000",
  "votes": 13,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi</p>\n<p>I am wondering what's the criteria for users who have bought more than 12 items in the test period: for example, are the first 12 items included, or the most frequently ones in case of repeated purchases, etc</p>\n<p>This is important when doing a time series split: for example, leaving out of the train data the last 7 days of transactions, and use those days to generate a ground truth which makes possible computing the MAP metric against this test set.</p>\n<p>Thanks!</p>",
  "messages": [
    {
      "id": 1687851,
      "postDate": "2022-02-13T07:50:17.587Z",
      "content": "<p>Hi</p>\n<p>I am wondering what's the criteria for users who have bought more than 12 items in the test period: for example, are the first 12 items included, or the most frequently ones in case of repeated purchases, etc</p>\n<p>This is important when doing a time series split: for example, leaving out of the train data the last 7 days of transactions, and use those days to generate a ground truth which makes possible computing the MAP metric against this test set.</p>\n<p>Thanks!</p>",
      "rawMarkdown": "Hi\n\nI am wondering what's the criteria for users who have bought more than 12 items in the test period: for example, are the first 12 items included, or the most frequently ones in case of repeated purchases, etc\n\nThis is important when doing a time series split: for example, leaving out of the train data the last 7 days of transactions, and use those days to generate a ground truth which makes possible computing the MAP metric against this test set.\n\nThanks!",
      "votes": 12
    },
    {
      "id": 1688863,
      "postDate": "2022-02-13T21:26:52.707Z",
      "content": "<p>I think that if any customer has more then 12 purchases in the test period that does n't matters the main thing is this that our 12 recommendations are how much precise or how much MAP they have .<br>\nHope that i am able to explain you <br>\nand this is only my opinion </p>",
      "rawMarkdown": "I think that if any customer has more then 12 purchases in the test period that does n't matters the main thing is this that our 12 recommendations are how much precise or how much MAP they have .\nHope that i am able to explain you \nand this is only my opinion ",
      "votes": 3,
      "isDeleted": true,
      "replies": [
        {
          "id": 1691103,
          "postDate": "2022-02-15T08:33:24.237Z",
          "content": "<p>Yes, I understand what you say and I think you may be right, it makes total sense: the customer can have more than 12 purchases; you predict up to 12 and each of your predictions is scored based on whether it has been purchased or not.</p>",
          "rawMarkdown": "Yes, I understand what you say and I think you may be right, it makes total sense: the customer can have more than 12 purchases; you predict up to 12 and each of your predictions is scored based on whether it has been purchased or not.",
          "votes": 2
        },
        {
          "id": 1691270,
          "postDate": "2022-02-15T10:21:56.363Z",
          "content": "<p>Now i am 100% sure that this is correct becuase it's confirmed by the kaggle staff</p>",
          "rawMarkdown": "Now i am 100% sure that this is correct becuase it's confirmed by the kaggle staff",
          "isDeleted": true
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 1688863,
      "author_name": "",
      "author_url": "",
      "post_date": "2022-02-13T21:26:52.707000",
      "content": "<p>I think that if any customer has more then 12 purchases in the test period that does n't matters the main thing is this that our 12 recommendations are how much precise or how much MAP they have .<br>\nHope that i am able to explain you <br>\nand this is only my opinion </p>",
      "votes": 3,
      "replies": [
        {
          "id": 1691103,
          "author_name": "Victor Gonzalez",
          "author_url": "",
          "post_date": "2022-02-15T08:33:24.237000",
          "content": "<p>Yes, I understand what you say and I think you may be right, it makes total sense: the customer can have more than 12 purchases; you predict up to 12 and each of your predictions is scored based on whether it has been purchased or not.</p>",
          "votes": 2,
          "replies": []
        },
        {
          "id": 1691270,
          "author_name": "",
          "author_url": "",
          "post_date": "2022-02-15T10:21:56.363000",
          "content": "<p>Now i am 100% sure that this is correct becuase it's confirmed by the kaggle staff</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1687851": "Hi\n\nI am wondering what's the criteria for users who have bought more than 12 items in the test period: for example, are the first 12 items included, or the most frequently ones in case of repeated purchases, etc\n\nThis is important when doing a time series split: for example, leaving out of the train data the last 7 days of transactions, and use those days to generate a ground truth which makes possible computing the MAP metric against this test set.\n\nThanks!",
    "1688863": "I think that if any customer has more then 12 purchases in the test period that does n't matters the main thing is this that our 12 recommendations are how much precise or how much MAP they have .\nHope that i am able to explain you \nand this is only my opinion "
  }
}