Here are the two article talking about the meaning of srch-destination-id and hotel_market, but both of these seem like the city name.\nCan anyone tell the difference between these two? Thanks!
\n\nHere are the two article talking about the meaning of srch-destination-id and hotel_market, but both of these seem like the city name.\nCan anyone tell the difference between these two? Thanks!
\n\nHotel_market divides the globe into non-overlapping regions such as: "Now York", "Boston", "Washington, D.C.", etc.
\n\nOn the other hand, srch-destination-id are nodes in a hierarchical taxonomy of overlapping search destinations. Examples of search destinations that correspond to the "New York" hotel_market are "New York and vicinity", "New York City", "JFK Airport", "Manhattan", "Statue of Liberty", "Central Park", etc.
\n\nThe taxonomy itself (e.g. search destination A is a "child" of search destination B) was not released here.
", "rawMarkdown": "Hotel_market divides the globe into non-overlapping regions such as: \"Now York\", \"Boston\", \"Washington, D.C.\", etc. \r\n\r\nOn the other hand, srch-destination-id are nodes in a hierarchical taxonomy of overlapping search destinations. Examples of search destinations that correspond to the \"New York\" hotel_market are \"New York and vicinity\", \"New York City\", \"JFK Airport\", \"Manhattan\", \"Statue of Liberty\", \"Central Park\", etc. \r\n\r\nThe taxonomy itself (e.g. search destination A is a \"child\" of search destination B) was not released here.", "votes": null }, { "id": "116609", "postDate": "04/25/2016 11:51:30", "content": "Adam,
\n\nyou said that srch-destination-id could overlap. \nCould you clarify, if a particular hotel belongs to two overlapping destinations will it be assigned to the same cluster for the both destinations?
\n\nMore generally - are clusters defined for each search destinations independently?
\nOr maybe clusters are defined on hotel market level?
Clustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.
", "rawMarkdown": "Clustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.", "votes": null }, { "id": "116614", "postDate": "04/25/2016 12:14:34", "content": "Thanks Adam.
\n\nSo basically a hotel_destination_id will always have the same hotel_market, but a hotel_market may have multiple destination_ids. That's what I observed in a subset of data and aligns well with what you say.
", "rawMarkdown": "Thanks Adam.\r\n\r\nSo basically a hotel_destination_id will always have the same hotel_market, but a hotel_market may have multiple destination_ids. That's what I observed in a *subset* of data and aligns well with what you say.", "votes": null }, { "id": "116615", "postDate": "04/25/2016 12:18:49", "content": "There might be in fact search destinations spanning several markets (e.g. "South of France") but from what you say these are not included in the data.
", "rawMarkdown": "There might be in fact search destinations spanning several markets (e.g. \"South of France\") but from what you say these are not included in the data.", "votes": null }, { "id": "116630", "postDate": "04/25/2016 13:37:44", "content": "[quote=Adam;116610]
\n\nClustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.
\n\n[/quote]
\n\nIt looks like this competition is about finding what the users values most, and less about geographical characteristics of the hotels. In that sense, for example, cluster #1 would be something like "inexpensive, pets allowed, free parking, within x miles from search_id,...", while cluster #2 would be "renovated, beachfront, inside a mall, within y miles from search_id...", etc. \nTo what extent this is correct? Thanks!
", "rawMarkdown": "[quote=Adam;116610]\r\n\r\nClustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.\r\n\r\n[/quote]\r\n\r\nIt looks like this competition is about finding what the users values most, and less about geographical characteristics of the hotels. In that sense, for example, cluster #1 would be something like \"inexpensive, pets allowed, free parking, within x miles from search_id,...\", while cluster #2 would be \"renovated, beachfront, inside a mall, within y miles from search_id...\", etc. \r\nTo what extent this is correct? Thanks!", "votes": null }, { "id": "116632", "postDate": "04/25/2016 13:43:38", "content": "Geographical characteristics are important because e.g. distance to region center is used in clustering. Otherwise, your interpretation is correct.
", "rawMarkdown": "Geographical characteristics are important because e.g. distance to region center is used in clustering. Otherwise, your interpretation is correct.", "votes": null }, { "id": "116638", "postDate": "04/25/2016 14:20:20", "content": "Yes, thanks Adam, by 'geographical' I wanted to say a geography in absolute terms, like Tokyo, Rio or London. And I see that x miles from search_destination_id is also a geographical characteristic, but relative to a search. That is its about what the user wants, not what the hotel is.
", "rawMarkdown": "Yes, thanks Adam, by 'geographical' I wanted to say a geography in absolute terms, like Tokyo, Rio or London. And I see that x miles from search_destination_id is also a geographical characteristic, but relative to a search. That is its about what the user wants, not what the hotel is.", "votes": null }, { "id": "117924", "postDate": "05/01/2016 19:24:34", "content": "Some hotel_market values span multiple hotel_country values. E.g. hotel_market 1774 is tied to hotel_country 9 and hotel_country 204. Is this a data error or should hotel_market only be used within the context of hotel_country? (e.g. hotel_market 1774 = San Antonio, hotel_country 9 = USA and hotel_country 204 = Argentina)
", "rawMarkdown": "Some hotel_market values span multiple hotel_country values. E.g. hotel_market 1774 is tied to hotel_country 9 and hotel_country 204. Is this a data error or should hotel_market only be used within the context of hotel_country? (e.g. hotel_market 1774 = San Antonio, hotel_country 9 = USA and hotel_country 204 = Argentina)", "votes": null }, { "id": "117935", "postDate": "05/01/2016 20:24:13", "content": "There are markets spanning multiple countries. It can be also a data error.
", "rawMarkdown": "There are markets spanning multiple countries. It can be also a data error.", "votes": null } ], "comments": [ { "id": 116580, "author_name": "adamwoz", "author_url": "", "post_date": "04/25/2016 09:09:17", "content": "Hotel_market divides the globe into non-overlapping regions such as: "Now York", "Boston", "Washington, D.C.", etc.
\n\nOn the other hand, srch-destination-id are nodes in a hierarchical taxonomy of overlapping search destinations. Examples of search destinations that correspond to the "New York" hotel_market are "New York and vicinity", "New York City", "JFK Airport", "Manhattan", "Statue of Liberty", "Central Park", etc.
\n\nThe taxonomy itself (e.g. search destination A is a "child" of search destination B) was not released here.
", "votes": null, "replies": [] }, { "id": 116609, "author_name": "anmiko", "author_url": "", "post_date": "04/25/2016 11:51:30", "content": "Adam,
\n\nyou said that srch-destination-id could overlap. \nCould you clarify, if a particular hotel belongs to two overlapping destinations will it be assigned to the same cluster for the both destinations?
\n\nMore generally - are clusters defined for each search destinations independently?
\nOr maybe clusters are defined on hotel market level?
Clustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.
", "votes": null, "replies": [] }, { "id": 116614, "author_name": "sachintyagi", "author_url": "", "post_date": "04/25/2016 12:14:34", "content": "Thanks Adam.
\n\nSo basically a hotel_destination_id will always have the same hotel_market, but a hotel_market may have multiple destination_ids. That's what I observed in a subset of data and aligns well with what you say.
", "votes": null, "replies": [] }, { "id": 116615, "author_name": "adamwoz", "author_url": "", "post_date": "04/25/2016 12:18:49", "content": "There might be in fact search destinations spanning several markets (e.g. "South of France") but from what you say these are not included in the data.
", "votes": null, "replies": [] }, { "id": 116630, "author_name": "hitoshinagano", "author_url": "", "post_date": "04/25/2016 13:37:44", "content": "[quote=Adam;116610]
\n\nClustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.
\n\n[/quote]
\n\nIt looks like this competition is about finding what the users values most, and less about geographical characteristics of the hotels. In that sense, for example, cluster #1 would be something like "inexpensive, pets allowed, free parking, within x miles from search_id,...", while cluster #2 would be "renovated, beachfront, inside a mall, within y miles from search_id...", etc. \nTo what extent this is correct? Thanks!
", "votes": null, "replies": [] }, { "id": 116632, "author_name": "adamwoz", "author_url": "", "post_date": "04/25/2016 13:43:38", "content": "Geographical characteristics are important because e.g. distance to region center is used in clustering. Otherwise, your interpretation is correct.
", "votes": null, "replies": [] }, { "id": 116638, "author_name": "hitoshinagano", "author_url": "", "post_date": "04/25/2016 14:20:20", "content": "Yes, thanks Adam, by 'geographical' I wanted to say a geography in absolute terms, like Tokyo, Rio or London. And I see that x miles from search_destination_id is also a geographical characteristic, but relative to a search. That is its about what the user wants, not what the hotel is.
", "votes": null, "replies": [] }, { "id": 117924, "author_name": "ben519", "author_url": "", "post_date": "05/01/2016 19:24:34", "content": "Some hotel_market values span multiple hotel_country values. E.g. hotel_market 1774 is tied to hotel_country 9 and hotel_country 204. Is this a data error or should hotel_market only be used within the context of hotel_country? (e.g. hotel_market 1774 = San Antonio, hotel_country 9 = USA and hotel_country 204 = Argentina)
", "votes": null, "replies": [] }, { "id": 117935, "author_name": "adamwoz", "author_url": "", "post_date": "05/01/2016 20:24:13", "content": "There are markets spanning multiple countries. It can be also a data error.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "116559": "Here are the two article talking about the meaning of srch-destination-id and hotel_market, but both of these seem like the city name.\r\nCan anyone tell the difference between these two? Thanks!\r\n\r\n1. srch-destination-id\r\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/forums/t/20286/what-does-srch-destination-id-mean/116010\r\n2. hotel_market\r\nhttps://www.kaggle.com/c/expedia-hotel-recommendations/forums/t/20169/what-does-hotel-market-exactly-mean/116441", "116580": "Hotel_market divides the globe into non-overlapping regions such as: \"Now York\", \"Boston\", \"Washington, D.C.\", etc. \r\n\r\nOn the other hand, srch-destination-id are nodes in a hierarchical taxonomy of overlapping search destinations. Examples of search destinations that correspond to the \"New York\" hotel_market are \"New York and vicinity\", \"New York City\", \"JFK Airport\", \"Manhattan\", \"Statue of Liberty\", \"Central Park\", etc. \r\n\r\nThe taxonomy itself (e.g. search destination A is a \"child\" of search destination B) was not released here.", "116609": "Adam,\r\n\r\nyou said that srch-destination-id could overlap. \r\nCould you clarify, if a particular hotel belongs to two overlapping destinations will it be assigned to the same cluster for the both destinations?\r\n\r\nMore generally - are clusters defined for each search destinations independently? \r\nOr maybe clusters are defined on hotel market level?", "116610": "Clustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.", "116614": "Thanks Adam.\r\n\r\nSo basically a hotel_destination_id will always have the same hotel_market, but a hotel_market may have multiple destination_ids. That's what I observed in a *subset* of data and aligns well with what you say.", "116615": "There might be in fact search destinations spanning several markets (e.g. \"South of France\") but from what you say these are not included in the data.", "116630": "[quote=Adam;116610]\r\n\r\nClustering was done globally and not per market / search destination. Since distance of a hotel to a center of a destination is used in clustering then it can happen that a hotel H is in cluster A in the context of search destination X and in cluster B in the context of search destination Y.\r\n\r\n[/quote]\r\n\r\nIt looks like this competition is about finding what the users values most, and less about geographical characteristics of the hotels. In that sense, for example, cluster #1 would be something like \"inexpensive, pets allowed, free parking, within x miles from search_id,...\", while cluster #2 would be \"renovated, beachfront, inside a mall, within y miles from search_id...\", etc. \r\nTo what extent this is correct? Thanks!", "116632": "Geographical characteristics are important because e.g. distance to region center is used in clustering. Otherwise, your interpretation is correct.", "116638": "Yes, thanks Adam, by 'geographical' I wanted to say a geography in absolute terms, like Tokyo, Rio or London. And I see that x miles from search_destination_id is also a geographical characteristic, but relative to a search. That is its about what the user wants, not what the hotel is.", "117924": "Some hotel_market values span multiple hotel_country values. E.g. hotel_market 1774 is tied to hotel_country 9 and hotel_country 204. Is this a data error or should hotel_market only be used within the context of hotel_country? (e.g. hotel_market 1774 = San Antonio, hotel_country 9 = USA and hotel_country 204 = Argentina)", "117935": "There are markets spanning multiple countries. It can be also a data error." }, "source": "meta" }