{
  "id": 66544,
  "title": "Use of data in research paper",
  "url": "/competitions/ga-customer-revenue-prediction/discussion/66544",
  "author_name": "",
  "post_date": "2018-09-22T12:29:27.606575200Z",
  "votes": 3,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Can this data be used in a research paper? Or are there any restrictions to it? </p>",
  "messages": [
    {
      "id": "391785",
      "postDate": "09/22/2018 12:29:27",
      "content": "<p>Can this data be used in a research paper? Or are there any restrictions to it? </p>",
      "rawMarkdown": "Can this data be used in a research paper? Or are there any restrictions to it?",
      "votes": null
    },
    {
      "id": "391928",
      "postDate": "09/22/2018 17:48:36",
      "content": "<p>Disclaimer: I am not a lawyer, though I sometimes play one on the internet.</p>\n\n<p>but this is how I read it:</p>\n\n<p><a href=\"https://www.kaggle.com/c/ga-customer-revenue-prediction/rules\">https://www.kaggle.com/c/ga-customer-revenue-prediction/rules</a>  section B.7.A says the data can be used for non-commercial academic research.  But you cannot release the data (section B.7.B,  presumably in part or in whole) in said paper, just general statistics and conclusions.  So no quoting dataframe rows, I'd say, but means and standard deviations of columns, should be ok.  As for graphs? should be ok if they don't allow one to reconstruct too much of the data.  My guess is the main concern is you don't give enough away to de-anonymize the data (as happened in a Netflix competition several years ago).</p>",
      "rawMarkdown": "Disclaimer: I am not a lawyer, though I sometimes play one on the internet.\n\nbut this is how I read it:\n\nhttps://www.kaggle.com/c/ga-customer-revenue-prediction/rules  section B.7.A says the data can be used for non-commercial academic research.  But you cannot release the data (section B.7.B,  presumably in part or in whole) in said paper, just general statistics and conclusions.  So no quoting dataframe rows, I'd say, but means and standard deviations of columns, should be ok.  As for graphs? should be ok if they don't allow one to reconstruct too much of the data.  My guess is the main concern is you don't give enough away to de-anonymize the data (as happened in a Netflix competition several years ago).",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 391928,
      "author_name": "depmountaineer",
      "author_url": "",
      "post_date": "09/22/2018 17:48:36",
      "content": "<p>Disclaimer: I am not a lawyer, though I sometimes play one on the internet.</p>\n\n<p>but this is how I read it:</p>\n\n<p><a href=\"https://www.kaggle.com/c/ga-customer-revenue-prediction/rules\">https://www.kaggle.com/c/ga-customer-revenue-prediction/rules</a>  section B.7.A says the data can be used for non-commercial academic research.  But you cannot release the data (section B.7.B,  presumably in part or in whole) in said paper, just general statistics and conclusions.  So no quoting dataframe rows, I'd say, but means and standard deviations of columns, should be ok.  As for graphs? should be ok if they don't allow one to reconstruct too much of the data.  My guess is the main concern is you don't give enough away to de-anonymize the data (as happened in a Netflix competition several years ago).</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "391785": "Can this data be used in a research paper? Or are there any restrictions to it?",
    "391928": "Disclaimer: I am not a lawyer, though I sometimes play one on the internet.\n\nbut this is how I read it:\n\nhttps://www.kaggle.com/c/ga-customer-revenue-prediction/rules  section B.7.A says the data can be used for non-commercial academic research.  But you cannot release the data (section B.7.B,  presumably in part or in whole) in said paper, just general statistics and conclusions.  So no quoting dataframe rows, I'd say, but means and standard deviations of columns, should be ok.  As for graphs? should be ok if they don't allow one to reconstruct too much of the data.  My guess is the main concern is you don't give enough away to de-anonymize the data (as happened in a Netflix competition several years ago)."
  },
  "source": "meta"
}