{
  "id": 102013,
  "title": "What is your least favorite molecule?",
  "url": "/competitions/champs-scalar-coupling/discussion/102013",
  "author_name": "somnambWl",
  "post_date": "2019-07-30T12:16:59.896000",
  "votes": 1,
  "comment_count": 2,
  "views": 0,
  "content": "<p>I was just looking at a scatter plot of predictions of the scalar coupling constant for 1JHN, looking at outliers when this question came to my mind.</p>\n\n<p>Do you have a least favourite molecule? Maybe an outlier with high error and you just can not make your model to predict it correctly?</p>\n\n<p>As for me, I keep a list of weird molecules and outliers I find. Now, for 1JHN it is the molecule 128113 where the error of scalar coupling constant is 18.12. I also believe that I have features included that should help the model to predict it correctly, but still, it remains an outlier.</p>",
  "messages": [
    {
      "id": 588313,
      "postDate": "2019-07-30T12:16:59.897Z",
      "content": "<p>I was just looking at a scatter plot of predictions of the scalar coupling constant for 1JHN, looking at outliers when this question came to my mind.</p>\n\n<p>Do you have a least favourite molecule? Maybe an outlier with high error and you just can not make your model to predict it correctly?</p>\n\n<p>As for me, I keep a list of weird molecules and outliers I find. Now, for 1JHN it is the molecule 128113 where the error of scalar coupling constant is 18.12. I also believe that I have features included that should help the model to predict it correctly, but still, it remains an outlier.</p>",
      "rawMarkdown": "I was just looking at a scatter plot of predictions of the scalar coupling constant for 1JHN, looking at outliers when this question came to my mind.\n\nDo you have a least favourite molecule? Maybe an outlier with high error and you just can not make your model to predict it correctly?\n\nAs for me, I keep a list of weird molecules and outliers I find. Now, for 1JHN it is the molecule 128113 where the error of scalar coupling constant is 18.12. I also believe that I have features included that should help the model to predict it correctly, but still, it remains an outlier.",
      "votes": 1
    },
    {
      "id": 590924,
      "postDate": "2019-08-02T20:38:47.603Z",
      "content": "<p>Hmm figured since we almost the same place on the leader board that we would have the same ballpark error - 128113 is 6.877354 putting it just inside our worst 25 list.</p>",
      "rawMarkdown": "Hmm figured since we almost the same place on the leader board that we would have the same ballpark error - 128113 is 6.877354 putting it just inside our worst 25 list."
    },
    {
      "id": 592725,
      "postDate": "2019-08-05T18:21:06.770Z",
      "rawMarkdown": "",
      "votes": -1,
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 590924,
      "author_name": "PC Jimmmy",
      "author_url": "",
      "post_date": "2019-08-02T20:38:47.603000",
      "content": "<p>Hmm figured since we almost the same place on the leader board that we would have the same ballpark error - 128113 is 6.877354 putting it just inside our worst 25 list.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 592725,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-08-05T18:21:06.770000",
      "content": "",
      "votes": -1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "588313": "I was just looking at a scatter plot of predictions of the scalar coupling constant for 1JHN, looking at outliers when this question came to my mind.\n\nDo you have a least favourite molecule? Maybe an outlier with high error and you just can not make your model to predict it correctly?\n\nAs for me, I keep a list of weird molecules and outliers I find. Now, for 1JHN it is the molecule 128113 where the error of scalar coupling constant is 18.12. I also believe that I have features included that should help the model to predict it correctly, but still, it remains an outlier.",
    "590924": "Hmm figured since we almost the same place on the leader board that we would have the same ballpark error - 128113 is 6.877354 putting it just inside our worst 25 list.",
    "592725": ""
  }
}