{
  "id": 73533,
  "title": "Question about MAP@5",
  "url": "/competitions/humpback-whale-identification/discussion/73533",
  "author_name": "",
  "post_date": "2018-12-03T21:28:40.893829Z",
  "votes": 2,
  "comment_count": 2,
  "views": 0,
  "content": "<p>Suppose for an given example, the ground truth is whale #1. Your model top picks are #1, #2, #3, #4, and #5. So the precision would be 1/1 + 1/2 +1/3 +1/4 + 1/5. However, if your model is super sure that it is just #1 (say 99% probability), does it make sense to submit #1, #1, #1, #1, and #1? The precision in this case would be 1/1 +1/1 +1/1 +1/1 +1/1, better than spreading your bet? or does Kaggle remove duplicate or cap it? Thanks :D</p>",
  "messages": [
    {
      "id": "432433",
      "postDate": "12/03/2018 21:28:40",
      "content": "<p>Suppose for an given example, the ground truth is whale #1. Your model top picks are #1, #2, #3, #4, and #5. So the precision would be 1/1 + 1/2 +1/3 +1/4 + 1/5. However, if your model is super sure that it is just #1 (say 99% probability), does it make sense to submit #1, #1, #1, #1, and #1? The precision in this case would be 1/1 +1/1 +1/1 +1/1 +1/1, better than spreading your bet? or does Kaggle remove duplicate or cap it? Thanks :D</p>",
      "rawMarkdown": "Suppose for an given example, the ground truth is whale #1. Your model top picks are #1, #2, #3, #4, and #5. So the precision would be 1/1 + 1/2 +1/3 +1/4 + 1/5. However, if your model is super sure that it is just #1 (say 99% probability), does it make sense to submit #1, #1, #1, #1, and #1? The precision in this case would be 1/1 +1/1 +1/1 +1/1 +1/1, better than spreading your bet? or does Kaggle remove duplicate or cap it? Thanks :D",
      "votes": null
    },
    {
      "id": "432476",
      "postDate": "12/03/2018 23:33:33",
      "content": "<p>Hey <a href=\"/zhenlanwang\">@zhenlanwang</a>,\nread <a href=\"https://www.kaggle.com/c/humpback-whale-identification/discussion/73303\">this discussion</a> and you can check <a href=\"https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric\">this kernel</a> too</p>",
      "rawMarkdown": "Hey @zhenlanwang,\nread [this discussion](https://www.kaggle.com/c/humpback-whale-identification/discussion/73303) and you can check [this kernel](https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric) too",
      "votes": null
    },
    {
      "id": "432497",
      "postDate": "12/04/2018 00:29:35",
      "content": "<p>Got it. Thank you for your explanation.</p>",
      "rawMarkdown": "Got it. Thank you for your explanation.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 432476,
      "author_name": "pestipeti",
      "author_url": "",
      "post_date": "12/03/2018 23:33:33",
      "content": "<p>Hey <a href=\"/zhenlanwang\">@zhenlanwang</a>,\nread <a href=\"https://www.kaggle.com/c/humpback-whale-identification/discussion/73303\">this discussion</a> and you can check <a href=\"https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric\">this kernel</a> too</p>",
      "votes": null,
      "replies": [
        {
          "id": 432497,
          "author_name": "zhenlanwang",
          "author_url": "",
          "post_date": "12/04/2018 00:29:35",
          "content": "<p>Got it. Thank you for your explanation.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "432433": "Suppose for an given example, the ground truth is whale #1. Your model top picks are #1, #2, #3, #4, and #5. So the precision would be 1/1 + 1/2 +1/3 +1/4 + 1/5. However, if your model is super sure that it is just #1 (say 99% probability), does it make sense to submit #1, #1, #1, #1, and #1? The precision in this case would be 1/1 +1/1 +1/1 +1/1 +1/1, better than spreading your bet? or does Kaggle remove duplicate or cap it? Thanks :D",
    "432476": "Hey @zhenlanwang,\nread [this discussion](https://www.kaggle.com/c/humpback-whale-identification/discussion/73303) and you can check [this kernel](https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric) too",
    "432497": "Got it. Thank you for your explanation."
  },
  "source": "meta"
}