{
  "id": 4165,
  "title": "Expected score of the winning solution?",
  "url": "/competitions/cause-effect-pairs/discussion/4165",
  "author_name": "",
  "post_date": "2013-04-02T06:20:44.597Z",
  "votes": null,
  "comment_count": 1,
  "views": 1576,
  "content": "<p>Dear Isabelle,</p>\r\n<p>first of all, thank you for preparing and hosting this competition!</p>\r\n<p>As you are already familiar with relevant research on identifiying causal dependencies purely from data it would be interesting if you could share some of&nbsp;your expectations for this competition. What do you think will be the score of the winning solution?</p>\r\n<p>It seems a reasonable base line solution would have a score around 0.75? The largest fraction of the training set are pairs of two numeric variables, around half of them are independent pairs, and from the remaining dependent pairs, half are A-&gt;B and half\r\n B-&gt;A.&nbsp;</p>",
  "messages": [
    {
      "id": "21977",
      "postDate": "04/02/2013 06:20:44",
      "content": "<p>Dear Isabelle,</p>\r\n<p>first of all, thank you for preparing and hosting this competition!</p>\r\n<p>As you are already familiar with relevant research on identifiying causal dependencies purely from data it would be interesting if you could share some of&nbsp;your expectations for this competition. What do you think will be the score of the winning solution?</p>\r\n<p>It seems a reasonable base line solution would have a score around 0.75? The largest fraction of the training set are pairs of two numeric variables, around half of them are independent pairs, and from the remaining dependent pairs, half are A-&gt;B and half\r\n B-&gt;A.&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "22012",
      "postDate": "04/02/2013 18:13:46",
      "content": "<p>I need to think deeply before I answer your question because I do not want to bias the competition, but I beloeve there is still a bit margin for improvement because some state-of-the-art methods miss cases that are obvious to the human eye, like non invertible\r\n functions. Interestingly, some cases are easy for authomatic methods and hard for the human eye.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 22012,
      "author_name": "iguyon",
      "author_url": "",
      "post_date": "04/02/2013 18:13:46",
      "content": "<p>I need to think deeply before I answer your question because I do not want to bias the competition, but I beloeve there is still a bit margin for improvement because some state-of-the-art methods miss cases that are obvious to the human eye, like non invertible\r\n functions. Interestingly, some cases are easy for authomatic methods and hard for the human eye.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "21977": "",
    "22012": ""
  },
  "source": "meta"
}