{
  "id": 13129,
  "title": "negative prob? LB rescore?",
  "url": "/competitions/malware-classification/discussion/13129",
  "author_name": "",
  "post_date": "2015-03-29T23:58:16.523Z",
  "votes": 2,
  "comment_count": 4,
  "views": 1031,
  "content": "<p>I just realized one of our previous submission has negative probabilities, like what happened in otto. However, new submissions with negative prob are not allowed. So are our submissions going to be re-scored? like otto?&nbsp;</p>",
  "messages": [
    {
      "id": "68933",
      "postDate": "03/29/2015 23:58:16",
      "content": "<p>I just realized one of our previous submission has negative probabilities, like what happened in otto. However, new submissions with negative prob are not allowed. So are our submissions going to be re-scored? like otto?&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "68937",
      "postDate": "03/30/2015 00:06:17",
      "content": "<p>Yes, they're already on the queue. We have 20 scoring instances working on it.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "68983",
      "postDate": "03/30/2015 06:03:46",
      "content": "<p>Is re-scoring on this one over?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "68985",
      "postDate": "03/30/2015 06:26:19",
      "content": "<p>Just out of curiosity, how do you end up with negative probabilities? Is that the result of recalibrating the output of your model?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "69026",
      "postDate": "03/30/2015 15:19:57",
      "content": "<p>@Abhishek, I think it's done. all our solutions with negative probs are invalidated.</p>\n<p>@ David, yeah it's a mistake when we try to calibrate predictions.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 68937,
      "author_name": "wcukierski",
      "author_url": "",
      "post_date": "03/30/2015 00:06:17",
      "content": "<p>Yes, they're already on the queue. We have 20 scoring instances working on it.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 68983,
      "author_name": "abhishek",
      "author_url": "",
      "post_date": "03/30/2015 06:03:46",
      "content": "<p>Is re-scoring on this one over?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 68985,
      "author_name": "davidtran",
      "author_url": "",
      "post_date": "03/30/2015 06:26:19",
      "content": "<p>Just out of curiosity, how do you end up with negative probabilities? Is that the result of recalibrating the output of your model?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 69026,
      "author_name": "jiweiliu",
      "author_url": "",
      "post_date": "03/30/2015 15:19:57",
      "content": "<p>@Abhishek, I think it's done. all our solutions with negative probs are invalidated.</p>\n<p>@ David, yeah it's a mistake when we try to calibrate predictions.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "68933": "",
    "68937": "",
    "68983": "",
    "68985": "",
    "69026": ""
  },
  "source": "meta"
}