{
  "id": 310189,
  "title": "evaluation metrics",
  "url": "/competitions/happy-whale-and-dolphin/discussion/310189",
  "author_name": "",
  "post_date": "2022-02-28T03:12:41.559121Z",
  "votes": 4,
  "comment_count": 1,
  "views": 0,
  "content": "<p>How to use mean average precision at k for classification task .<br>\nIf i am not wrong , mean average precision at k is for object detection and this is classification task then why evaluation metrics for comp is this one, and how to implement for classification.<br>\nand while calculating mean average precision at k we first calculate precision at k in which we say <br>\n<code>y_pred = y_pred[:k]</code>  but why there is k because in multiclass classification one image will have one target and one prediction then why k because while calculating average precision at 5 we calculate precision at 1 to 5 and average them but if image has only one target then why to calculate for 5 times and take average it will be same </p>\n<p>is this multiclass classification problem or object detection problem?</p>",
  "messages": [
    {
      "id": "1706981",
      "postDate": "02/28/2022 03:12:41",
      "content": "<p>How to use mean average precision at k for classification task .<br>\nIf i am not wrong , mean average precision at k is for object detection and this is classification task then why evaluation metrics for comp is this one, and how to implement for classification.<br>\nand while calculating mean average precision at k we first calculate precision at k in which we say <br>\n<code>y_pred = y_pred[:k]</code>  but why there is k because in multiclass classification one image will have one target and one prediction then why k because while calculating average precision at 5 we calculate precision at 1 to 5 and average them but if image has only one target then why to calculate for 5 times and take average it will be same </p>\n<p>is this multiclass classification problem or object detection problem?</p>",
      "rawMarkdown": "How to use mean average precision at k for classification task .\nIf i am not wrong , mean average precision at k is for object detection and this is classification task then why evaluation metrics for comp is this one, and how to implement for classification.\nand while calculating mean average precision at k we first calculate precision at k in which we say \n```y_pred = y_pred[:k]```  but why there is k because in multiclass classification one image will have one target and one prediction then why k because while calculating average precision at 5 we calculate precision at 1 to 5 and average them but if image has only one target then why to calculate for 5 times and take average it will be same \n\nis this multiclass classification problem or object detection problem?",
      "votes": null
    },
    {
      "id": "1707957",
      "postDate": "03/01/2022 01:06:09",
      "content": "<p>It confused me at first too, since the MAP score is so common in object detection. I thought this notebook gave a very clear explanation: <a href=\"https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric\" target=\"_blank\">https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric</a></p>\n<p>Best of luck!</p>",
      "rawMarkdown": "It confused me at first too, since the MAP score is so common in object detection. I thought this notebook gave a very clear explanation: [https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric](https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric)\n\nBest of luck!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1707957,
      "author_name": "rturley",
      "author_url": "",
      "post_date": "03/01/2022 01:06:09",
      "content": "<p>It confused me at first too, since the MAP score is so common in object detection. I thought this notebook gave a very clear explanation: <a href=\"https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric\" target=\"_blank\">https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric</a></p>\n<p>Best of luck!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1706981": "How to use mean average precision at k for classification task .\nIf i am not wrong , mean average precision at k is for object detection and this is classification task then why evaluation metrics for comp is this one, and how to implement for classification.\nand while calculating mean average precision at k we first calculate precision at k in which we say \n```y_pred = y_pred[:k]```  but why there is k because in multiclass classification one image will have one target and one prediction then why k because while calculating average precision at 5 we calculate precision at 1 to 5 and average them but if image has only one target then why to calculate for 5 times and take average it will be same \n\nis this multiclass classification problem or object detection problem?",
    "1707957": "It confused me at first too, since the MAP score is so common in object detection. I thought this notebook gave a very clear explanation: [https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric](https://www.kaggle.com/pestipeti/explanation-of-map5-scoring-metric)\n\nBest of luck!"
  },
  "source": "meta"
}