{
  "id": 10814,
  "title": "About the Open Source requirement",
  "url": "/competitions/avazu-ctr-prediction/discussion/10814",
  "author_name": "",
  "post_date": "2014-11-01T12:08:56.780Z",
  "votes": null,
  "comment_count": 2,
  "views": 1254,
  "content": "<p>I have a question regarding the Open Source requirement for the code.</p>\n\n<p><code>A Submission will be ineligible to win a prize if it was developed using code containing or depending on software licensed under an open source license:</code><code>* other than an Open Source Initiative-approved license (see &lt;http://opensource.org/&gt;); or<br>* an open source license that prohibits commercial use.</code></p>\n\n<p>If I put the data in a SQL Server database (which is not open source) to gather&nbsp;some statistics about the data and perform queries.&nbsp;</p>\n<p>And then use some C# or R code<strong> to&nbsp;train a model using the&nbsp;CSV file</strong>&nbsp;will this make my submission ineligible to win a prize?<br><br></p>\n<p>I wonder because in this case the whole training and prediction code would be open source, but the first step of my methodology would not be reproductible by someone not having SQL Server.&nbsp;<br><br>So should I use other databases for this step too or is it ok to use SQL Server?</p>",
  "messages": [
    {
      "id": "57160",
      "postDate": "11/01/2014 12:08:56",
      "content": "<p>I have a question regarding the Open Source requirement for the code.</p>\n\n<p><code>A Submission will be ineligible to win a prize if it was developed using code containing or depending on software licensed under an open source license:</code><code>* other than an Open Source Initiative-approved license (see &lt;http://opensource.org/&gt;); or<br>* an open source license that prohibits commercial use.</code></p>\n\n<p>If I put the data in a SQL Server database (which is not open source) to gather&nbsp;some statistics about the data and perform queries.&nbsp;</p>\n<p>And then use some C# or R code<strong> to&nbsp;train a model using the&nbsp;CSV file</strong>&nbsp;will this make my submission ineligible to win a prize?<br><br></p>\n<p>I wonder because in this case the whole training and prediction code would be open source, but the first step of my methodology would not be reproductible by someone not having SQL Server.&nbsp;<br><br>So should I use other databases for this step too or is it ok to use SQL Server?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "57162",
      "postDate": "11/01/2014 12:37:23",
      "content": "<p>using database to do sql like processing is accepted</p>\n<p>using database embedded data mining tools (as I know, most of them are commercial and gui based) won't win a prize</p>\n<p>to be&nbsp;reproducible, you can also provide the pre-processed csv file</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "57170",
      "postDate": "11/01/2014 18:26:53",
      "content": "<p>I can recommend the R package RMySQL if you like to query your data frames in SQL but want a pure R solution. &nbsp;Don't know if you're worried about query performance/memory requirement though that would push towards a database?</p>\n<p>EDIT: I meant to say the R package sqldf, but either might be appropriate here!</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 57162,
      "author_name": "stevewang0",
      "author_url": "",
      "post_date": "11/01/2014 12:37:23",
      "content": "<p>using database to do sql like processing is accepted</p>\n<p>using database embedded data mining tools (as I know, most of them are commercial and gui based) won't win a prize</p>\n<p>to be&nbsp;reproducible, you can also provide the pre-processed csv file</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 57170,
      "author_name": "jaymoore",
      "author_url": "",
      "post_date": "11/01/2014 18:26:53",
      "content": "<p>I can recommend the R package RMySQL if you like to query your data frames in SQL but want a pure R solution. &nbsp;Don't know if you're worried about query performance/memory requirement though that would push towards a database?</p>\n<p>EDIT: I meant to say the R package sqldf, but either might be appropriate here!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "57160": "",
    "57162": "",
    "57170": ""
  },
  "source": "meta"
}