{
  "id": 192344,
  "title": " You should not try to submit anything for the rows that contain lectures",
  "url": "/competitions/riiid-test-answer-prediction/discussion/192344",
  "author_name": "",
  "post_date": "2020-10-21T04:41:04.755162200Z",
  "votes": null,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Hi </p>\n<p>A s per  the   data  section </p>\n<p>You should not try to submit anything for the rows that contain lectures.</p>\n<p>So what should be done   if we encounter  rows that  contain lectures <br>\n1) should  we  delete  those rows?<br>\n2)Should we  skip  those rows while making predictions?<br>\n3) Will the  test data have  any such  rows in  the  first place ?</p>\n<p>Somebody  please  calrify?</p>",
  "messages": [
    {
      "id": "1055747",
      "postDate": "10/21/2020 04:41:04",
      "content": "<p>Hi </p>\n<p>A s per  the   data  section </p>\n<p>You should not try to submit anything for the rows that contain lectures.</p>\n<p>So what should be done   if we encounter  rows that  contain lectures <br>\n1) should  we  delete  those rows?<br>\n2)Should we  skip  those rows while making predictions?<br>\n3) Will the  test data have  any such  rows in  the  first place ?</p>\n<p>Somebody  please  calrify?</p>",
      "rawMarkdown": "Hi \n\nA s per  the   data  section \n\nYou should not try to submit anything for the rows that contain lectures.\n\nSo what should be done   if we encounter  rows that  contain lectures \n1) should  we  delete  those rows?\n2)Should we  skip  those rows while making predictions?\n3) Will the  test data have  any such  rows in  the  first place ?\n\nSomebody  please  calrify?",
      "votes": null
    },
    {
      "id": "1055749",
      "postDate": "10/21/2020 04:43:26",
      "content": "<p>I am not sure why people feel that it's fine to keep creating topics which have been discussed multiple times and specified clearly? I would really suggest to take a deep dive at the old discussions first…</p>\n<p>You shouldn't make preds on those rows.</p>",
      "rawMarkdown": "I am not sure why people feel that it's fine to keep creating topics which have been discussed multiple times and specified clearly? I would really suggest to take a deep dive at the old discussions first...\n\nYou shouldn't make preds on those rows.",
      "votes": null
    },
    {
      "id": "1056436",
      "postDate": "10/21/2020 17:40:28",
      "content": "<p>Hi, I’m new to Kaggle and I’ve stumbled onto small mistakes like this as well. Let’s be nice to each other. You’re probably right but your tone is discouraging towards beginners. </p>",
      "rawMarkdown": "Hi, I’m new to Kaggle and I’ve stumbled onto small mistakes like this as well. Let’s be nice to each other. You’re probably right but your tone is discouraging towards beginners.",
      "votes": null
    },
    {
      "id": "1056450",
      "postDate": "10/21/2020 17:54:24",
      "content": "<p>Well I didn't mean to hurt anyone's sentiments, my sincere apologies. I was also new once but i kept quiet for almost ~1.5 years, just observed people, looked for answers, used to read a lot here. So after spending 4+ years on kaggle, i can definitely say that there's plethora of information on the dataset as to what to do and what not to do, many kernel's have shown successful subs, multiple helpful posts, etc. This comp style new and the same is for everyone. People are happy reading whole data in 10 secs but by doing that they forget that they have to first understand what's there in the data and one can do that by taking just 1M rows pretty much IMHO (to get started, the imp. of whole data comes at a later stage, right?), create features on that and see how they look like, look for possibility to get nan's etc. And then there's data DESC, multiple discussion posts where people have tried sharing what one should be careful with as well. </p>\n<p>The above is what i am also doing right now as well.</p>\n<p>Apologies Again.</p>",
      "rawMarkdown": "Well I didn't mean to hurt anyone's sentiments, my sincere apologies. I was also new once but i kept quiet for almost ~1.5 years, just observed people, looked for answers, used to read a lot here. So after spending 4+ years on kaggle, i can definitely say that there's plethora of information on the dataset as to what to do and what not to do, many kernel's have shown successful subs, multiple helpful posts, etc. This comp style new and the same is for everyone. People are happy reading whole data in 10 secs but by doing that they forget that they have to first understand what's there in the data and one can do that by taking just 1M rows pretty much IMHO (to get started, the imp. of whole data comes at a later stage, right?), create features on that and see how they look like, look for possibility to get nan's etc. And then there's data DESC, multiple discussion posts where people have tried sharing what one should be careful with as well. \n\nThe above is what i am also doing right now as well.\n\nApologies Again.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1055749,
      "author_name": "adityaecdrid",
      "author_url": "",
      "post_date": "10/21/2020 04:43:26",
      "content": "<p>I am not sure why people feel that it's fine to keep creating topics which have been discussed multiple times and specified clearly? I would really suggest to take a deep dive at the old discussions first…</p>\n<p>You shouldn't make preds on those rows.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1056436,
          "author_name": "dinodelao",
          "author_url": "",
          "post_date": "10/21/2020 17:40:28",
          "content": "<p>Hi, I’m new to Kaggle and I’ve stumbled onto small mistakes like this as well. Let’s be nice to each other. You’re probably right but your tone is discouraging towards beginners. </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1056450,
          "author_name": "adityaecdrid",
          "author_url": "",
          "post_date": "10/21/2020 17:54:24",
          "content": "<p>Well I didn't mean to hurt anyone's sentiments, my sincere apologies. I was also new once but i kept quiet for almost ~1.5 years, just observed people, looked for answers, used to read a lot here. So after spending 4+ years on kaggle, i can definitely say that there's plethora of information on the dataset as to what to do and what not to do, many kernel's have shown successful subs, multiple helpful posts, etc. This comp style new and the same is for everyone. People are happy reading whole data in 10 secs but by doing that they forget that they have to first understand what's there in the data and one can do that by taking just 1M rows pretty much IMHO (to get started, the imp. of whole data comes at a later stage, right?), create features on that and see how they look like, look for possibility to get nan's etc. And then there's data DESC, multiple discussion posts where people have tried sharing what one should be careful with as well. </p>\n<p>The above is what i am also doing right now as well.</p>\n<p>Apologies Again.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1055747": "Hi \n\nA s per  the   data  section \n\nYou should not try to submit anything for the rows that contain lectures.\n\nSo what should be done   if we encounter  rows that  contain lectures \n1) should  we  delete  those rows?\n2)Should we  skip  those rows while making predictions?\n3) Will the  test data have  any such  rows in  the  first place ?\n\nSomebody  please  calrify?",
    "1055749": "I am not sure why people feel that it's fine to keep creating topics which have been discussed multiple times and specified clearly? I would really suggest to take a deep dive at the old discussions first...\n\nYou shouldn't make preds on those rows.",
    "1056436": "Hi, I’m new to Kaggle and I’ve stumbled onto small mistakes like this as well. Let’s be nice to each other. You’re probably right but your tone is discouraging towards beginners.",
    "1056450": "Well I didn't mean to hurt anyone's sentiments, my sincere apologies. I was also new once but i kept quiet for almost ~1.5 years, just observed people, looked for answers, used to read a lot here. So after spending 4+ years on kaggle, i can definitely say that there's plethora of information on the dataset as to what to do and what not to do, many kernel's have shown successful subs, multiple helpful posts, etc. This comp style new and the same is for everyone. People are happy reading whole data in 10 secs but by doing that they forget that they have to first understand what's there in the data and one can do that by taking just 1M rows pretty much IMHO (to get started, the imp. of whole data comes at a later stage, right?), create features on that and see how they look like, look for possibility to get nan's etc. And then there's data DESC, multiple discussion posts where people have tried sharing what one should be careful with as well. \n\nThe above is what i am also doing right now as well.\n\nApologies Again."
  },
  "source": "meta"
}