{
  "id": 198434,
  "title": "What's the size of LB/PB hidden testset?",
  "url": "/competitions/riiid-test-answer-prediction/discussion/198434",
  "author_name": "",
  "post_date": "2020-11-21T07:52:56.309039400Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Refering to data section, it says \"Expect to see roughly 2.5 million questions in the hidden test set\".  <br>\nDoes it mean we have 2.5M rows of test data(LB+PB)? Maybe half of them splits into LB?</p>\n<p>I want to create my CV strategy. Our <code>train.csv</code> contains ~110M rows of data, which is quite huge compared to 2.5M test set. Did I get something wrong?</p>",
  "messages": [
    {
      "id": "1085795",
      "postDate": "11/21/2020 07:52:56",
      "content": "<p>Refering to data section, it says \"Expect to see roughly 2.5 million questions in the hidden test set\".  <br>\nDoes it mean we have 2.5M rows of test data(LB+PB)? Maybe half of them splits into LB?</p>\n<p>I want to create my CV strategy. Our <code>train.csv</code> contains ~110M rows of data, which is quite huge compared to 2.5M test set. Did I get something wrong?</p>",
      "rawMarkdown": "Refering to data section, it says \"Expect to see roughly 2.5 million questions in the hidden test set\".  \nDoes it mean we have 2.5M rows of test data(LB+PB)? Maybe half of them splits into LB?\n\nI want to create my CV strategy. Our `train.csv` contains ~110M rows of data, which is quite huge compared to 2.5M test set. Did I get something wrong?",
      "votes": null
    },
    {
      "id": "1086342",
      "postDate": "11/21/2020 15:54:41",
      "content": "<p>The leaderboard is calculated with approximately 20% of the test data, and there are 2.5 million questions in the test set. So, I guess the size of LB is about 0.5 million and the size of PB is about 2 million.</p>",
      "rawMarkdown": "The leaderboard is calculated with approximately 20% of the test data, and there are 2.5 million questions in the test set. So, I guess the size of LB is about 0.5 million and the size of PB is about 2 million.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1086342,
      "author_name": "mamasinkgs",
      "author_url": "",
      "post_date": "11/21/2020 15:54:41",
      "content": "<p>The leaderboard is calculated with approximately 20% of the test data, and there are 2.5 million questions in the test set. So, I guess the size of LB is about 0.5 million and the size of PB is about 2 million.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1085795": "Refering to data section, it says \"Expect to see roughly 2.5 million questions in the hidden test set\".  \nDoes it mean we have 2.5M rows of test data(LB+PB)? Maybe half of them splits into LB?\n\nI want to create my CV strategy. Our `train.csv` contains ~110M rows of data, which is quite huge compared to 2.5M test set. Did I get something wrong?",
    "1086342": "The leaderboard is calculated with approximately 20% of the test data, and there are 2.5 million questions in the test set. So, I guess the size of LB is about 0.5 million and the size of PB is about 2 million."
  },
  "source": "meta"
}