{
  "id": 175444,
  "title": "Percent columns",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/175444",
  "author_name": "",
  "post_date": "2020-08-18T07:46:35.917131Z",
  "votes": 2,
  "comment_count": 3,
  "views": 0,
  "content": "<p>Can someone help me understand the Percent column.<br>\nAnd as a lot of the competitior have used this notebook<br>\n<a href=\"url\" target=\"_blank\">https://www.kaggle.com/ulrich07/osic-multiple-quantile-regression-starter</a><br>\ncan anyone who has understood the approach in the notebook clearly  explain what has been done with the percent column for the submission set<br>\nThere it seems that the percent value has been considered same for all weeks</p>",
  "messages": [
    {
      "id": "975218",
      "postDate": "08/18/2020 07:46:35",
      "content": "<p>Can someone help me understand the Percent column.<br>\nAnd as a lot of the competitior have used this notebook<br>\n<a href=\"url\" target=\"_blank\">https://www.kaggle.com/ulrich07/osic-multiple-quantile-regression-starter</a><br>\ncan anyone who has understood the approach in the notebook clearly  explain what has been done with the percent column for the submission set<br>\nThere it seems that the percent value has been considered same for all weeks</p>",
      "rawMarkdown": "Can someone help me understand the Percent column.\nAnd as a lot of the competitior have used this notebook\n[https://www.kaggle.com/ulrich07/osic-multiple-quantile-regression-starter](url)\ncan anyone who has understood the approach in the notebook clearly  explain what has been done with the percent column for the submission set\nThere it seems that the percent value has been considered same for all weeks",
      "votes": null
    },
    {
      "id": "977306",
      "postDate": "08/19/2020 12:06:17",
      "content": "<p>I hope I understood the question correctly. If so, then the issue is that we are predicting the patient's FVC for multiple weeks which will not be in the training / test dataset. This means that the Percent value for each week is not available to us. Hence, people are utilizing aggregated Percent values and statistics for every patient and adding that to the submission prediction dataframes.</p>\n<p>**BTW, I tried using two models. 1st to predict percentage by Age,Sex, Smoking status, and past percentage values and then imported them to the second model for predicting FVC. My CV remained similar and LB was little worse. (Without percent prediction, linear regression : 7.1x, with percent prediction, linear regression : 7.0x)</p>\n<p>Cheers</p>",
      "rawMarkdown": "I hope I understood the question correctly. If so, then the issue is that we are predicting the patient's FVC for multiple weeks which will not be in the training / test dataset. This means that the Percent value for each week is not available to us. Hence, people are utilizing aggregated Percent values and statistics for every patient and adding that to the submission prediction dataframes.\n\n**BTW, I tried using two models. 1st to predict percentage by Age,Sex, Smoking status, and past percentage values and then imported them to the second model for predicting FVC. My CV remained similar and LB was little worse. (Without percent prediction, linear regression : 7.1x, with percent prediction, linear regression : 7.0x)\n\nCheers",
      "votes": null
    },
    {
      "id": "989547",
      "postDate": "08/28/2020 22:33:06",
      "content": "<p>Hi,<br>\nFirstly before talking about the Percent column, let's understand these two variables (<strong><em>FVC</em></strong> and <strong><em>FEV1</em></strong>  ):<br>\n-<strong>forced vital capacity(VFC)</strong>  measure  the volume  of air that can exhaled out of your lungs following a deep inhalation.<br>\n-<strong>Forced expiratory volume in 1 second (FEV1)</strong> measure  the volume  of air that can exhaled out of your lungs <br>\n  in one second following a deep inhalation.</p>\n<p>Now the <strong>Percent</strong> is a ratio between FVC and FEV1 (<strong>FEV1 /FVC</strong>) and it  represents the percent of the FVC that can be exhaled in one second.</p>\n<p>Hope it's clear.</p>",
      "rawMarkdown": "Hi,\nFirstly before talking about the Percent column, let's understand these two variables (***FVC*** and ***FEV1***  ):\n-**forced vital capacity(VFC)**  measure  the volume  of air that can exhaled out of your lungs following a deep inhalation.\n-**Forced expiratory volume in 1 second (FEV1)** measure  the volume  of air that can exhaled out of your lungs \n  in one second following a deep inhalation.\n\nNow the **Percent** is a ratio between FVC and FEV1 (**FEV1 /FVC**) and it  represents the percent of the FVC that can be exhaled in one second.\n\nHope it's clear.",
      "votes": null
    },
    {
      "id": "1026088",
      "postDate": "09/25/2020 04:24:36",
      "content": "<p>Percent is not ratio of FEV1/FVC  ,it is  percent of some expected value for a given patients age,gender </p>",
      "rawMarkdown": "Percent is not ratio of FEV1/FVC  ,it is  percent of some expected value for a given patients age,gender",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 977306,
      "author_name": "aliasgherman",
      "author_url": "",
      "post_date": "08/19/2020 12:06:17",
      "content": "<p>I hope I understood the question correctly. If so, then the issue is that we are predicting the patient's FVC for multiple weeks which will not be in the training / test dataset. This means that the Percent value for each week is not available to us. Hence, people are utilizing aggregated Percent values and statistics for every patient and adding that to the submission prediction dataframes.</p>\n<p>**BTW, I tried using two models. 1st to predict percentage by Age,Sex, Smoking status, and past percentage values and then imported them to the second model for predicting FVC. My CV remained similar and LB was little worse. (Without percent prediction, linear regression : 7.1x, with percent prediction, linear regression : 7.0x)</p>\n<p>Cheers</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 989547,
      "author_name": "azzeineaftiss",
      "author_url": "",
      "post_date": "08/28/2020 22:33:06",
      "content": "<p>Hi,<br>\nFirstly before talking about the Percent column, let's understand these two variables (<strong><em>FVC</em></strong> and <strong><em>FEV1</em></strong>  ):<br>\n-<strong>forced vital capacity(VFC)</strong>  measure  the volume  of air that can exhaled out of your lungs following a deep inhalation.<br>\n-<strong>Forced expiratory volume in 1 second (FEV1)</strong> measure  the volume  of air that can exhaled out of your lungs <br>\n  in one second following a deep inhalation.</p>\n<p>Now the <strong>Percent</strong> is a ratio between FVC and FEV1 (<strong>FEV1 /FVC</strong>) and it  represents the percent of the FVC that can be exhaled in one second.</p>\n<p>Hope it's clear.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1026088,
          "author_name": "jaideepvalani",
          "author_url": "",
          "post_date": "09/25/2020 04:24:36",
          "content": "<p>Percent is not ratio of FEV1/FVC  ,it is  percent of some expected value for a given patients age,gender </p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "975218": "Can someone help me understand the Percent column.\nAnd as a lot of the competitior have used this notebook\n[https://www.kaggle.com/ulrich07/osic-multiple-quantile-regression-starter](url)\ncan anyone who has understood the approach in the notebook clearly  explain what has been done with the percent column for the submission set\nThere it seems that the percent value has been considered same for all weeks",
    "977306": "I hope I understood the question correctly. If so, then the issue is that we are predicting the patient's FVC for multiple weeks which will not be in the training / test dataset. This means that the Percent value for each week is not available to us. Hence, people are utilizing aggregated Percent values and statistics for every patient and adding that to the submission prediction dataframes.\n\n**BTW, I tried using two models. 1st to predict percentage by Age,Sex, Smoking status, and past percentage values and then imported them to the second model for predicting FVC. My CV remained similar and LB was little worse. (Without percent prediction, linear regression : 7.1x, with percent prediction, linear regression : 7.0x)\n\nCheers",
    "989547": "Hi,\nFirstly before talking about the Percent column, let's understand these two variables (***FVC*** and ***FEV1***  ):\n-**forced vital capacity(VFC)**  measure  the volume  of air that can exhaled out of your lungs following a deep inhalation.\n-**Forced expiratory volume in 1 second (FEV1)** measure  the volume  of air that can exhaled out of your lungs \n  in one second following a deep inhalation.\n\nNow the **Percent** is a ratio between FVC and FEV1 (**FEV1 /FVC**) and it  represents the percent of the FVC that can be exhaled in one second.\n\nHope it's clear.",
    "1026088": "Percent is not ratio of FEV1/FVC  ,it is  percent of some expected value for a given patients age,gender"
  },
  "source": "meta"
}