{
  "id": 5750,
  "title": "Customer ID?",
  "url": "/competitions/expedia-personalized-sort/discussion/5750",
  "author_name": "",
  "post_date": "2013-09-13T16:07:56.713Z",
  "votes": null,
  "comment_count": 4,
  "views": 1496,
  "content": "<p>Hi!</p>\n<p>I was about to start this competition and I noticed that the dataset does not provide explicit customer IDs tracked through a sign-in or a cookie. I could however notice that the tracking is being done, as the customer's historic average of rating is available.</p>\n<p>Am I missing a column here or was the customer ID not provided on purpose?</p>\n<p>Sid</p>",
  "messages": [
    {
      "id": "30861",
      "postDate": "09/13/2013 16:07:56",
      "content": "<p>Hi!</p>\n<p>I was about to start this competition and I noticed that the dataset does not provide explicit customer IDs tracked through a sign-in or a cookie. I could however notice that the tracking is being done, as the customer's historic average of rating is available.</p>\n<p>Am I missing a column here or was the customer ID not provided on purpose?</p>\n<p>Sid</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "30863",
      "postDate": "09/13/2013 16:12:10",
      "content": "<p>This column was not provided in this contest on purpose.</p>\n<p>Adam</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "30865",
      "postDate": "09/13/2013 16:23:50",
      "content": "<p>Thanks for the quick reply. I can understand some reasons why this column is omitted.&nbsp;Absence of this column eliminates all conventional collaborative filtering algorithms except for a few. Which is fine because most of the conventional ones would not have performed good anyway because of the sparsity of the data.</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31011",
      "postDate": "09/16/2013 21:29:55",
      "content": "<p>I totally agree with your point Siddharta, we would not have more insight about the booking habits, however we would probably have more information on the click habits, that are not reflected in the visitor_hist_starrating and the visitor_hist_adr_usd. &nbsp;</p>\n<p>I would like to know more about the reason why there is no costumer ID, because obviously, the better we know the people that we want to predict the actions, the better we predict their actions.&nbsp;Is it a legal issue ?&nbsp;Or do you think it introduces a bias, in favor of the current ordering ? (The first hotels you provide would be more clicked, but only because they are first on the list, and not because they fit the costumers expectations better.)</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "31025",
      "postDate": "09/17/2013 07:26:36",
      "content": "<p>We decided not to include user id mainly due to privacy concerns.</p>\n<p>&nbsp;</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 30863,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "09/13/2013 16:12:10",
      "content": "<p>This column was not provided in this contest on purpose.</p>\n<p>Adam</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 30865,
      "author_name": "",
      "author_url": "",
      "post_date": "09/13/2013 16:23:50",
      "content": "<p>Thanks for the quick reply. I can understand some reasons why this column is omitted.&nbsp;Absence of this column eliminates all conventional collaborative filtering algorithms except for a few. Which is fine because most of the conventional ones would not have performed good anyway because of the sparsity of the data.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31011,
      "author_name": "florentrenucci",
      "author_url": "",
      "post_date": "09/16/2013 21:29:55",
      "content": "<p>I totally agree with your point Siddharta, we would not have more insight about the booking habits, however we would probably have more information on the click habits, that are not reflected in the visitor_hist_starrating and the visitor_hist_adr_usd. &nbsp;</p>\n<p>I would like to know more about the reason why there is no costumer ID, because obviously, the better we know the people that we want to predict the actions, the better we predict their actions.&nbsp;Is it a legal issue ?&nbsp;Or do you think it introduces a bias, in favor of the current ordering ? (The first hotels you provide would be more clicked, but only because they are first on the list, and not because they fit the costumers expectations better.)</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 31025,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "09/17/2013 07:26:36",
      "content": "<p>We decided not to include user id mainly due to privacy concerns.</p>\n<p>&nbsp;</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "30861": "",
    "30863": "",
    "30865": "",
    "31011": "",
    "31025": ""
  },
  "source": "meta"
}