{
  "id": 3578,
  "title": "Final event recommendations",
  "url": "/competitions/event-recommendation-engine-challenge/discussion/3578",
  "author_name": "",
  "post_date": "2013-01-12T00:00:28.693Z",
  "votes": null,
  "comment_count": 3,
  "views": 2067,
  "content": "Based off the benchmark code, do we just need rank events that are in test set for that user? Also, user event pair in test set will not come up in event-attendees right? Asking questions to understand the problem. Thanks in advance.",
  "messages": [
    {
      "id": "19208",
      "postDate": "01/12/2013 00:00:28",
      "content": "Based off the benchmark code, do we just need rank events that are in test set for that user? Also, user event pair in test set will not come up in event-attendees right? Asking questions to understand the problem. Thanks in advance.",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19209",
      "postDate": "01/12/2013 00:17:19",
      "content": "<p>[quote=Triton-SD;19208]</p>\r\n<p>Based off the benchmark code, do we just need rank events that are in test set for that user? Also, user event pair in test set will not come up in event-attendees right? Asking questions to understand the problem. Thanks in advance.</p>\r\n<p>[/quote]Yes, both of those are correct.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19274",
      "postDate": "01/13/2013 22:45:33",
      "content": "<p>This thread has thrown me off.&nbsp; What is said on the evaluation page:</p>\r\n<blockquote>\r\n<p>For each user in the test set, you are submitting a list of events. These events should be ordered from the ones in which you predict the user will be most interested to those in which the user will be least be interested.</p>\r\n</blockquote>\r\n<p>implies that I can recommend any events to the user that I predict he/she will be interested in.</p>\r\n<p>However Triton-SD's comments and your response suggest that the task is only to\r\n<span style=\"text-decoration:underline\">order</span> the events already present for that user in the test set.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "19275",
      "postDate": "01/14/2013 00:21:46",
      "content": "<p>Yes, I am also a bit confused.</p>\r\n<p>I looked at the 2 benchmarks. The random one has the events from the test.csv, ordered in some way.</p>\r\n<p>However, the event_popularity benchmark has events that were <em>NOT</em> in test.csv. Is this a valid answer, or was it just given as example of the evaluation metric, but the real answer SHOULD have only the events already present for this user_id in test.csv\r\n ?</p>\r\n<p>best regards</p>\r\n<p>Nikolay</p>\r\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 19209,
      "author_name": "benhamner",
      "author_url": "",
      "post_date": "01/12/2013 00:17:19",
      "content": "<p>[quote=Triton-SD;19208]</p>\r\n<p>Based off the benchmark code, do we just need rank events that are in test set for that user? Also, user event pair in test set will not come up in event-attendees right? Asking questions to understand the problem. Thanks in advance.</p>\r\n<p>[/quote]Yes, both of those are correct.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19274,
      "author_name": "lovi9573",
      "author_url": "",
      "post_date": "01/13/2013 22:45:33",
      "content": "<p>This thread has thrown me off.&nbsp; What is said on the evaluation page:</p>\r\n<blockquote>\r\n<p>For each user in the test set, you are submitting a list of events. These events should be ordered from the ones in which you predict the user will be most interested to those in which the user will be least be interested.</p>\r\n</blockquote>\r\n<p>implies that I can recommend any events to the user that I predict he/she will be interested in.</p>\r\n<p>However Triton-SD's comments and your response suggest that the task is only to\r\n<span style=\"text-decoration:underline\">order</span> the events already present for that user in the test set.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 19275,
      "author_name": "dieselboy",
      "author_url": "",
      "post_date": "01/14/2013 00:21:46",
      "content": "<p>Yes, I am also a bit confused.</p>\r\n<p>I looked at the 2 benchmarks. The random one has the events from the test.csv, ordered in some way.</p>\r\n<p>However, the event_popularity benchmark has events that were <em>NOT</em> in test.csv. Is this a valid answer, or was it just given as example of the evaluation metric, but the real answer SHOULD have only the events already present for this user_id in test.csv\r\n ?</p>\r\n<p>best regards</p>\r\n<p>Nikolay</p>\r\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "19208": "",
    "19209": "",
    "19274": "",
    "19275": ""
  },
  "source": "meta"
}