{
  "id": 21589,
  "title": "Ideas/suggestions to help Expedia?",
  "url": "/competitions/expedia-hotel-recommendations/discussion/21589",
  "author_name": "Goncalo Gordo",
  "post_date": "2016-06-11T07:23:20.913000",
  "votes": 0,
  "comment_count": 0,
  "views": 914,
  "content": "<p>I'm curious to learn about ideas other people might have that could be useful for Expedia.</p>\n\n<p>Here are mine:</p>\n\n<p>1) To improve algorithm performance (share more data with Kagglers)</p>\n\n<ul>\n<li>More data on users (to help understand their preferences): which hotels were shown to users on the search that led them to click or book the hotel; and how they chose to sort by hotels in the search results page (price, guest rating, hotel class, etc.)</li>\n<li>More data on hotels (to help understand why a user clicked/booked it): e.g. guest ratings, average room price, size (# of rooms), occupancy</li>\n<li>Less hotel clusters (less overlap between clusters): some more experienced Kagglers mentioned it was hard making classification algs work and Adam also mentioned some overlap/fuzziness in clusters definition (maybe a potential new exercise/competition would be providing hotel data so that the community could also suggest a new alternative for the hotel clusters</li>\n</ul>\n\n<p>2) To improve their business (share more data with clients)</p>\n\n<ul>\n<li>Provide the conversion rate to hotels: so that hotels could tailor their offering to try to improve their conversion rates</li>\n<li>Provide info on &quot;lost&quot; clients to hotels: tell them about users that clicked on their hotel but ended up booking other hotels, and suggest potential actions they could take based on differences to hotels that ended up being the chosen ones (e.g. if the booked hotels were similar to theirs but cheaper, they could suggest lowering the prices; other options would be adding certain services, adding kids facilities, improving customer service, etc.)</li>\n</ul>\n\n<p>What else could they do?</p>",
  "messages": [
    {
      "id": 123346,
      "postDate": "2016-06-11T07:23:20.913Z",
      "content": "<p>I'm curious to learn about ideas other people might have that could be useful for Expedia.</p>\n\n<p>Here are mine:</p>\n\n<p>1) To improve algorithm performance (share more data with Kagglers)</p>\n\n<ul>\n<li>More data on users (to help understand their preferences): which hotels were shown to users on the search that led them to click or book the hotel; and how they chose to sort by hotels in the search results page (price, guest rating, hotel class, etc.)</li>\n<li>More data on hotels (to help understand why a user clicked/booked it): e.g. guest ratings, average room price, size (# of rooms), occupancy</li>\n<li>Less hotel clusters (less overlap between clusters): some more experienced Kagglers mentioned it was hard making classification algs work and Adam also mentioned some overlap/fuzziness in clusters definition (maybe a potential new exercise/competition would be providing hotel data so that the community could also suggest a new alternative for the hotel clusters</li>\n</ul>\n\n<p>2) To improve their business (share more data with clients)</p>\n\n<ul>\n<li>Provide the conversion rate to hotels: so that hotels could tailor their offering to try to improve their conversion rates</li>\n<li>Provide info on &quot;lost&quot; clients to hotels: tell them about users that clicked on their hotel but ended up booking other hotels, and suggest potential actions they could take based on differences to hotels that ended up being the chosen ones (e.g. if the booked hotels were similar to theirs but cheaper, they could suggest lowering the prices; other options would be adding certain services, adding kids facilities, improving customer service, etc.)</li>\n</ul>\n\n<p>What else could they do?</p>",
      "rawMarkdown": "I'm curious to learn about ideas other people might have that could be useful for Expedia.\r\n\r\nHere are mine:\r\n\r\n1) To improve algorithm performance (share more data with Kagglers)\r\n\r\n- More data on users (to help understand their preferences): which hotels were shown to users on the search that led them to click or book the hotel; and how they chose to sort by hotels in the search results page (price, guest rating, hotel class, etc.)\r\n- More data on hotels (to help understand why a user clicked/booked it): e.g. guest ratings, average room price, size (# of rooms), occupancy\r\n- Less hotel clusters (less overlap between clusters): some more experienced Kagglers mentioned it was hard making classification algs work and Adam also mentioned some overlap/fuzziness in clusters definition (maybe a potential new exercise/competition would be providing hotel data so that the community could also suggest a new alternative for the hotel clusters\r\n\r\n\r\n2) To improve their business (share more data with clients)\r\n\r\n - Provide the conversion rate to hotels: so that hotels could tailor their offering to try to improve their conversion rates\r\n - Provide info on \"lost\" clients to hotels: tell them about users that clicked on their hotel but ended up booking other hotels, and suggest potential actions they could take based on differences to hotels that ended up being the chosen ones (e.g. if the booked hotels were similar to theirs but cheaper, they could suggest lowering the prices; other options would be adding certain services, adding kids facilities, improving customer service, etc.)\r\n\r\nWhat else could they do?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "123346": "I'm curious to learn about ideas other people might have that could be useful for Expedia.\r\n\r\nHere are mine:\r\n\r\n1) To improve algorithm performance (share more data with Kagglers)\r\n\r\n- More data on users (to help understand their preferences): which hotels were shown to users on the search that led them to click or book the hotel; and how they chose to sort by hotels in the search results page (price, guest rating, hotel class, etc.)\r\n- More data on hotels (to help understand why a user clicked/booked it): e.g. guest ratings, average room price, size (# of rooms), occupancy\r\n- Less hotel clusters (less overlap between clusters): some more experienced Kagglers mentioned it was hard making classification algs work and Adam also mentioned some overlap/fuzziness in clusters definition (maybe a potential new exercise/competition would be providing hotel data so that the community could also suggest a new alternative for the hotel clusters\r\n\r\n\r\n2) To improve their business (share more data with clients)\r\n\r\n - Provide the conversion rate to hotels: so that hotels could tailor their offering to try to improve their conversion rates\r\n - Provide info on \"lost\" clients to hotels: tell them about users that clicked on their hotel but ended up booking other hotels, and suggest potential actions they could take based on differences to hotels that ended up being the chosen ones (e.g. if the booked hotels were similar to theirs but cheaper, they could suggest lowering the prices; other options would be adding certain services, adding kids facilities, improving customer service, etc.)\r\n\r\nWhat else could they do?"
  }
}