{
  "id": 21261,
  "title": "Collaborative filtering",
  "url": "/competitions/expedia-hotel-recommendations/discussion/21261",
  "author_name": "",
  "post_date": "2016-05-27T07:41:49.473Z",
  "votes": null,
  "comment_count": 1,
  "views": 475,
  "content": "<p>Hello, </p>\n\n<p>I was considering entering the competion. Since, I am currently working on collaborative filtering techniques, I was thinking about a user-rating matrix with items as hotel clusters. And give it a first shot using neighboorhods methods. </p>\n\n<p>Has anyone tried this approach? Is this feasible? </p>\n\n<p>Thanks for you help! </p>",
  "messages": [
    {
      "id": "121546",
      "postDate": "05/27/2016 07:41:49",
      "content": "<p>Hello, </p>\n\n<p>I was considering entering the competion. Since, I am currently working on collaborative filtering techniques, I was thinking about a user-rating matrix with items as hotel clusters. And give it a first shot using neighboorhods methods. </p>\n\n<p>Has anyone tried this approach? Is this feasible? </p>\n\n<p>Thanks for you help! </p>",
      "rawMarkdown": "Hello, \r\n\r\nI was considering entering the competion. Since, I am currently working on collaborative filtering techniques, I was thinking about a user-rating matrix with items as hotel clusters. And give it a first shot using neighboorhods methods. \r\n\r\nHas anyone tried this approach? Is this feasible? \r\n\r\nThanks for you help!",
      "votes": null
    },
    {
      "id": "121548",
      "postDate": "05/27/2016 07:45:17",
      "content": "<p>Hi,</p>\n\n<p>I tried the exact same approach using spark ALS,\nThe performance I got were correct, but lower than with other approaches.\nHowever, I am not an expert of collaborative filtering, so it might probably be improved.\n(In particular I suspect that a SVD technique with both product and features in, might give better results)</p>\n\n<p>Loic</p>",
      "rawMarkdown": "Hi,\r\n\r\nI tried the exact same approach using spark ALS,\r\nThe performance I got were correct, but lower than with other approaches.\r\nHowever, I am not an expert of collaborative filtering, so it might probably be improved.\r\n(In particular I suspect that a SVD technique with both product and features in, might give better results)\r\n\r\nLoic",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 121548,
      "author_name": "loicus",
      "author_url": "",
      "post_date": "05/27/2016 07:45:17",
      "content": "<p>Hi,</p>\n\n<p>I tried the exact same approach using spark ALS,\nThe performance I got were correct, but lower than with other approaches.\nHowever, I am not an expert of collaborative filtering, so it might probably be improved.\n(In particular I suspect that a SVD technique with both product and features in, might give better results)</p>\n\n<p>Loic</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "121546": "Hello, \r\n\r\nI was considering entering the competion. Since, I am currently working on collaborative filtering techniques, I was thinking about a user-rating matrix with items as hotel clusters. And give it a first shot using neighboorhods methods. \r\n\r\nHas anyone tried this approach? Is this feasible? \r\n\r\nThanks for you help!",
    "121548": "Hi,\r\n\r\nI tried the exact same approach using spark ALS,\r\nThe performance I got were correct, but lower than with other approaches.\r\nHowever, I am not an expert of collaborative filtering, so it might probably be improved.\r\n(In particular I suspect that a SVD technique with both product and features in, might give better results)\r\n\r\nLoic"
  },
  "source": "meta"
}