{
  "id": 20223,
  "title": "What does d1-d149 mean in destinations",
  "url": "/competitions/expedia-hotel-recommendations/discussion/20223",
  "author_name": "",
  "post_date": "2016-04-18T13:10:49.173Z",
  "votes": 4,
  "comment_count": 8,
  "views": 1813,
  "content": "<p>Hi,\nWhat does d1-d149 indicate in destinations file? \nRegards\nAiswarya</p>",
  "messages": [
    {
      "id": "115379",
      "postDate": "04/18/2016 13:10:49",
      "content": "<p>Hi,\nWhat does d1-d149 indicate in destinations file? \nRegards\nAiswarya</p>",
      "rawMarkdown": "Hi,\r\nWhat does d1-d149 indicate in destinations file? \r\nRegards\r\nAiswarya",
      "votes": null
    },
    {
      "id": "115460",
      "postDate": "04/18/2016 18:50:38",
      "content": "<p>This is a latent description of hotel reviews that are related to a given search destination. These columns correspond to different facets \n(e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. </p>\n\n<p>Adam</p>",
      "rawMarkdown": "This is a latent description of hotel reviews that are related to a given search destination. These columns correspond to different facets \r\n(e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \r\n\r\nAdam",
      "votes": null
    },
    {
      "id": "115884",
      "postDate": "04/20/2016 15:38:56",
      "content": "<p>hello,</p>\n\n<p>I am really confused with the interpretation of this file.</p>\n\n<ol>\n<li>Are these D1-D149 columns properties of the Destination?</li>\n<li>The user is already giving the destination in the search then how is the destination properties relevant?</li>\n<li>Is the destination file based on the training file? That is are the probabilities calculated from training file.\n(Users who endorsed/total users). </li>\n</ol>\n\n<p>regards\nGaurav</p>",
      "rawMarkdown": "hello,\r\n\r\nI am really confused with the interpretation of this file.\r\n\r\n1. Are these D1-D149 columns properties of the Destination?\r\n2. The user is already giving the destination in the search then how is the destination properties relevant?\r\n3. Is the destination file based on the training file? That is are the probabilities calculated from training file.\r\n(Users who endorsed/total users). \r\n\r\nregards\r\nGaurav",
      "votes": null
    },
    {
      "id": "116012",
      "postDate": "04/21/2016 13:00:16",
      "content": "<ol>\n<li>Yes</li>\n<li>You can think of it as extra properties of a destination. </li>\n<li>Information that was used to come up with the D1-D149 columns is not included in the training file. Keep in mind that D1-D149 are based on free text (users reviews). </li>\n</ol>",
      "rawMarkdown": "1. Yes\r\n2. You can think of it as extra properties of a destination. \r\n3. Information that was used to come up with the D1-D149 columns is not included in the training file. Keep in mind that D1-D149 are based on free text (users reviews).",
      "votes": null
    },
    {
      "id": "116037",
      "postDate": "04/21/2016 16:23:05",
      "content": "<p>Are D1-D149 TF-IDF features from Bag of Words?</p>",
      "rawMarkdown": "Are D1-D149 TF-IDF features from Bag of Words?",
      "votes": null
    },
    {
      "id": "116038",
      "postDate": "04/21/2016 16:24:03",
      "content": "<p>No</p>",
      "rawMarkdown": "No",
      "votes": null
    },
    {
      "id": "116047",
      "postDate": "04/21/2016 17:26:33",
      "content": "<p>It has been said in a couple places that these variable are latent dimensions. It sounds like they are SVD or word2vec or something from text reviews. Can I suggest that the admin state exactly how they are created  or say that it cant be exposed, in order to put it to rest?</p>",
      "rawMarkdown": "It has been said in a couple places that these variable are latent dimensions. It sounds like they are SVD or word2vec or something from text reviews. Can I suggest that the admin state exactly how they are created  or say that it cant be exposed, in order to put it to rest?",
      "votes": null
    },
    {
      "id": "116053",
      "postDate": "04/21/2016 17:45:01",
      "content": "<p>As I wrote above: &quot;These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. &quot;. Aspect based sentiment analysis was used to extract these values.</p>",
      "rawMarkdown": "As I wrote above: \"These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \". Aspect based sentiment analysis was used to extract these values.",
      "votes": null
    },
    {
      "id": "116055",
      "postDate": "04/21/2016 17:56:04",
      "content": "<p>[quote=Adam;116053]</p>\n\n<p>As I wrote above: &quot;These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. &quot;. Aspect based sentiment analysis was used to extract these values.</p>\n\n<p>[/quote]</p>\n\n<p>&quot;Aspect based sentiment analysis&quot; that helps! Thanks</p>",
      "rawMarkdown": "[quote=Adam;116053]\r\n\r\nAs I wrote above: \"These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \". Aspect based sentiment analysis was used to extract these values.\r\n\r\n\r\n\r\n[/quote]\r\n\r\n\r\n\"Aspect based sentiment analysis\" that helps! Thanks",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 115460,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "04/18/2016 18:50:38",
      "content": "<p>This is a latent description of hotel reviews that are related to a given search destination. These columns correspond to different facets \n(e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. </p>\n\n<p>Adam</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 115884,
      "author_name": "headliner",
      "author_url": "",
      "post_date": "04/20/2016 15:38:56",
      "content": "<p>hello,</p>\n\n<p>I am really confused with the interpretation of this file.</p>\n\n<ol>\n<li>Are these D1-D149 columns properties of the Destination?</li>\n<li>The user is already giving the destination in the search then how is the destination properties relevant?</li>\n<li>Is the destination file based on the training file? That is are the probabilities calculated from training file.\n(Users who endorsed/total users). </li>\n</ol>\n\n<p>regards\nGaurav</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116012,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "04/21/2016 13:00:16",
      "content": "<ol>\n<li>Yes</li>\n<li>You can think of it as extra properties of a destination. </li>\n<li>Information that was used to come up with the D1-D149 columns is not included in the training file. Keep in mind that D1-D149 are based on free text (users reviews). </li>\n</ol>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116037,
      "author_name": "thomasseleck",
      "author_url": "",
      "post_date": "04/21/2016 16:23:05",
      "content": "<p>Are D1-D149 TF-IDF features from Bag of Words?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116038,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "04/21/2016 16:24:03",
      "content": "<p>No</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116047,
      "author_name": "inspector",
      "author_url": "",
      "post_date": "04/21/2016 17:26:33",
      "content": "<p>It has been said in a couple places that these variable are latent dimensions. It sounds like they are SVD or word2vec or something from text reviews. Can I suggest that the admin state exactly how they are created  or say that it cant be exposed, in order to put it to rest?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116053,
      "author_name": "adamwoz",
      "author_url": "",
      "post_date": "04/21/2016 17:45:01",
      "content": "<p>As I wrote above: &quot;These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. &quot;. Aspect based sentiment analysis was used to extract these values.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 116055,
      "author_name": "inspector",
      "author_url": "",
      "post_date": "04/21/2016 17:56:04",
      "content": "<p>[quote=Adam;116053]</p>\n\n<p>As I wrote above: &quot;These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. &quot;. Aspect based sentiment analysis was used to extract these values.</p>\n\n<p>[/quote]</p>\n\n<p>&quot;Aspect based sentiment analysis&quot; that helps! Thanks</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "115379": "Hi,\r\nWhat does d1-d149 indicate in destinations file? \r\nRegards\r\nAiswarya",
    "115460": "This is a latent description of hotel reviews that are related to a given search destination. These columns correspond to different facets \r\n(e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \r\n\r\nAdam",
    "115884": "hello,\r\n\r\nI am really confused with the interpretation of this file.\r\n\r\n1. Are these D1-D149 columns properties of the Destination?\r\n2. The user is already giving the destination in the search then how is the destination properties relevant?\r\n3. Is the destination file based on the training file? That is are the probabilities calculated from training file.\r\n(Users who endorsed/total users). \r\n\r\nregards\r\nGaurav",
    "116012": "1. Yes\r\n2. You can think of it as extra properties of a destination. \r\n3. Information that was used to come up with the D1-D149 columns is not included in the training file. Keep in mind that D1-D149 are based on free text (users reviews).",
    "116037": "Are D1-D149 TF-IDF features from Bag of Words?",
    "116038": "No",
    "116047": "It has been said in a couple places that these variable are latent dimensions. It sounds like they are SVD or word2vec or something from text reviews. Can I suggest that the admin state exactly how they are created  or say that it cant be exposed, in order to put it to rest?",
    "116053": "As I wrote above: \"These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \". Aspect based sentiment analysis was used to extract these values.",
    "116055": "[quote=Adam;116053]\r\n\r\nAs I wrote above: \"These columns correspond to different facets (e.g. beach, ski, etc.) and values are (log) probabilities that a customer would endorse a hotel in the destination for a specific facet. \". Aspect based sentiment analysis was used to extract these values.\r\n\r\n\r\n\r\n[/quote]\r\n\r\n\r\n\"Aspect based sentiment analysis\" that helps! Thanks"
  },
  "source": "meta"
}