{
  "id": 49337,
  "title": "run-length encoding and threshold levels",
  "url": "/competitions/data-science-bowl-2018/discussion/49337",
  "author_name": "",
  "post_date": "2018-02-09T15:43:18.783448200Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, \nIn the evaluation section, it is stated that the IoU is calculated at different threshold levels. However, I wonder to know how it is derived from the submitted csv file since it is just consisted of  the positions and the length for each detected cells in the image. \nIn other word we have to set a threshold ourselves before creating the csv file for submission. Any suggestions?\nThanks!</p>",
  "messages": [
    {
      "id": "280234",
      "postDate": "02/09/2018 15:43:18",
      "content": "<p>Hi, \nIn the evaluation section, it is stated that the IoU is calculated at different threshold levels. However, I wonder to know how it is derived from the submitted csv file since it is just consisted of  the positions and the length for each detected cells in the image. \nIn other word we have to set a threshold ourselves before creating the csv file for submission. Any suggestions?\nThanks!</p>",
      "rawMarkdown": "Hi, \nIn the evaluation section, it is stated that the IoU is calculated at different threshold levels. However, I wonder to know how it is derived from the submitted csv file since it is just consisted of  the positions and the length for each detected cells in the image. \nIn other word we have to set a threshold ourselves before creating the csv file for submission. Any suggestions?\nThanks!",
      "votes": null
    },
    {
      "id": "280268",
      "postDate": "02/09/2018 17:12:36",
      "content": "<p>I think the meaning of thresholds is completely independent of how your generate your csv file for submission. In essence, it is only used to calculate the precision so you don't really need to worry about it unless you want to duplicate the evaluation metric for which there is a sample kernel here to help you: <a href=\"https://www.kaggle.com/wcukierski/example-metric-implementation\">https://www.kaggle.com/wcukierski/example-metric-implementation</a> . </p>",
      "rawMarkdown": "I think the meaning of thresholds is completely independent of how your generate your csv file for submission. In essence, it is only used to calculate the precision so you don't really need to worry about it unless you want to duplicate the evaluation metric for which there is a sample kernel here to help you: https://www.kaggle.com/wcukierski/example-metric-implementation .",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 280268,
      "author_name": "livingprogram",
      "author_url": "",
      "post_date": "02/09/2018 17:12:36",
      "content": "<p>I think the meaning of thresholds is completely independent of how your generate your csv file for submission. In essence, it is only used to calculate the precision so you don't really need to worry about it unless you want to duplicate the evaluation metric for which there is a sample kernel here to help you: <a href=\"https://www.kaggle.com/wcukierski/example-metric-implementation\">https://www.kaggle.com/wcukierski/example-metric-implementation</a> . </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "280234": "Hi, \nIn the evaluation section, it is stated that the IoU is calculated at different threshold levels. However, I wonder to know how it is derived from the submitted csv file since it is just consisted of  the positions and the length for each detected cells in the image. \nIn other word we have to set a threshold ourselves before creating the csv file for submission. Any suggestions?\nThanks!",
    "280268": "I think the meaning of thresholds is completely independent of how your generate your csv file for submission. In essence, it is only used to calculate the precision so you don't really need to worry about it unless you want to duplicate the evaluation metric for which there is a sample kernel here to help you: https://www.kaggle.com/wcukierski/example-metric-implementation ."
  },
  "source": "meta"
}