{
  "id": 306854,
  "title": "Why no negative samples?",
  "url": "/competitions/h-and-m-personalized-fashion-recommendations/discussion/306854",
  "author_name": "",
  "post_date": "2022-02-11T08:17:00.348263800Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>How can we do supervised recommendation without negative samples?</p>",
  "messages": [
    {
      "id": "1685381",
      "postDate": "02/11/2022 08:17:00",
      "content": "<p>How can we do supervised recommendation without negative samples?</p>",
      "rawMarkdown": "How can we do supervised recommendation without negative samples?",
      "votes": null
    },
    {
      "id": "1685739",
      "postDate": "02/11/2022 14:08:28",
      "content": "<p>Looks like you want to set this up as a binary classification question - will a customer buy an articles during the evaluation week.</p>\n<p>What's the training data, then?<br>\nProbably whether this/other customer/s bought this/other articles in previous weeks.<br>\nEach customer/article/week combination would be a single training example.</p>\n<p>Why aren't there negative examples, then?<br>\nThere are <strong><em>many, many more</em></strong> customer/article/week combinations where there was no sale, than where there was a sale.</p>",
      "rawMarkdown": "Looks like you want to set this up as a binary classification question - will a customer buy an articles during the evaluation week.\n\nWhat's the training data, then?\nProbably whether this/other customer/s bought this/other articles in previous weeks.\nEach customer/article/week combination would be a single training example.\n\nWhy aren't there negative examples, then?\nThere are ***many, many more*** customer/article/week combinations where there was no sale, than where there was a sale.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1685739,
      "author_name": "jacob34",
      "author_url": "",
      "post_date": "02/11/2022 14:08:28",
      "content": "<p>Looks like you want to set this up as a binary classification question - will a customer buy an articles during the evaluation week.</p>\n<p>What's the training data, then?<br>\nProbably whether this/other customer/s bought this/other articles in previous weeks.<br>\nEach customer/article/week combination would be a single training example.</p>\n<p>Why aren't there negative examples, then?<br>\nThere are <strong><em>many, many more</em></strong> customer/article/week combinations where there was no sale, than where there was a sale.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1685381": "How can we do supervised recommendation without negative samples?",
    "1685739": "Looks like you want to set this up as a binary classification question - will a customer buy an articles during the evaluation week.\n\nWhat's the training data, then?\nProbably whether this/other customer/s bought this/other articles in previous weeks.\nEach customer/article/week combination would be a single training example.\n\nWhy aren't there negative examples, then?\nThere are ***many, many more*** customer/article/week combinations where there was no sale, than where there was a sale."
  },
  "source": "meta"
}