![\"enter](\"https://media.giphy.com/media/Xxei1HDpBhMys/giphy.gif\")
Here we are, just a few days before the end, and it seems that there have not been any leaks in this competition. The way things stand right now, it looks like the team that has the best model and has gleaned the most from the data, will, in fact, win this thing. The organizer and Kaggle have done an impeccable job of preparing this dataset. This goes against all the expectations that we have from TalkingData competitions.
", "messages": [ { "id": 323281, "postDate": "2018-05-04T18:45:16.327Z", "content": "Here we are, just a few days before the end, and it seems that there have not been any leaks in this competition. The way things stand right now, it looks like the team that has the best model and has gleaned the most from the data, will, in fact, win this thing. The organizer and Kaggle have done an impeccable job of preparing this dataset. This goes against all the expectations that we have from TalkingData competitions.
", "rawMarkdown": "Here we are, just a few days before the end, and it seems that there have not been any leaks in this competition. The way things stand right now, it looks like the team that has the best model and has gleaned the most from the data, will, in fact, win this thing. The organizer and Kaggle have done an impeccable job of preparing this dataset. This goes against all the expectations that we have from TalkingData competitions. ", "votes": 23 }, { "id": 323287, "postDate": "2018-05-04T18:57:17.160Z", "content": "@Bojan - We'll do our best to make it up to you in future competitions. ;-)
", "rawMarkdown": "@Bojan - We'll do our best to make it up to you in future competitions. ;-)", "votes": 14, "replies": [ { "id": 323296, "postDate": "2018-05-04T19:21:12.713Z", "content": "Hope so. Would hate to see this become a trend.
", "rawMarkdown": "Hope so. Would hate to see this become a trend.", "votes": 7 } ] }, { "id": 323512, "postDate": "2018-05-05T10:17:46.367Z", "content": "You mean \"except IP%27 leak\"?
", "rawMarkdown": "You mean \"except IP%27 leak\"?", "votes": 8, "replies": [ { "id": 323528, "postDate": "2018-05-05T11:22:11.187Z", "content": "shhhhh, don't tell anyone!
", "rawMarkdown": "shhhhh, don't tell anyone!", "votes": 4 }, { "id": 323549, "postDate": "2018-05-05T13:04:18.570Z", "content": "Not again xD
", "rawMarkdown": "Not again xD", "votes": 3 }, { "id": 323724, "postDate": "2018-05-06T02:16:46.817Z", "content": "ip%27 leak??? serious?
", "rawMarkdown": "ip%27 leak??? serious?\n", "votes": 1 }, { "id": 323753, "postDate": "2018-05-06T05:07:13.770Z", "content": "@Peter it was a joke from another post
", "rawMarkdown": "@Peter it was a joke from another post", "votes": 2 } ] }, { "id": 325514, "postDate": "2018-05-08T13:43:52.953Z", "content": "Sorry to really disappoint you Bojan, but there was a (minor) leak: test data was sorted by click time AND is_attributed. Exploiting this could yield some boost, but nothing like previous major leaks we saw. In my case the boost was about 0.0004.
\n\nThe leak was publicly shared by @plantsgo soon enough for people to take advantage of it.
", "rawMarkdown": "Sorry to really disappoint you Bojan, but there was a (minor) leak: test data was sorted by click time AND is_attributed. Exploiting this could yield some boost, but nothing like previous major leaks we saw. In my case the boost was about 0.0004.\n\nThe leak was publicly shared by @plantsgo soon enough for people to take advantage of it.", "votes": 5, "replies": [ { "id": 326322, "postDate": "2018-05-09T15:10:47.743Z", "content": "Well played, very well played. I really should pay more attention to the discussion next time. There were several small tips and clues that could have made the difference. The problem with this competition was that the dataset was so large, it was not easy to iterate quickly and try many things, even the ones I was aware of.
", "rawMarkdown": "Well played, very well played. I really should pay more attention to the discussion next time. There were several small tips and clues that could have made the difference. The problem with this competition was that the dataset was so large, it was not easy to iterate quickly and try many things, even the ones I was aware of. ", "votes": 2 }, { "id": 326336, "postDate": "2018-05-09T15:25:02.183Z", "content": "I have been thinking of it for a month, see here, However, @plantsgo share made me try the one thing I was looking at, namely modifying the target in train.
\n\nThat said, I don't know if I would have tried it if he had not shared. As you say, there were so many things to try, and so little time, that prioritizing was difficult.
", "rawMarkdown": "I have been thinking of it for a month, see [here][1], However, @plantsgo share made me try the one thing I was looking at, namely modifying the target in train. \n\nThat said, I don't know if I would have tried it if he had not shared. As you say, there were so many things to try, and so little time, that prioritizing was difficult.\n\n [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/54229", "votes": 2 } ] }, { "id": 323290, "postDate": "2018-05-04T19:02:38.107Z", "content": "
This is a classic reference :-) Only a few upvotes Tilii, I wonder how many of many of us are old enough (and North American) to have watched it on TV?
", "rawMarkdown": "This is a classic reference :-) Only a few upvotes Tilii, I wonder how many of many of us are old enough (and North American) to have watched it on TV?", "votes": 2 }, { "id": 323716, "postDate": "2018-05-06T00:51:57.463Z", "content": "That actually is Tilii in the gif. ;-)
\n\nI’ve never watched it live on TV, not was I in the US at the time when it aired. I’ve only heard of it thanks to the idiom it lead to.
", "rawMarkdown": "That actually **is** Tilii in the gif. ;-)\n\nI’ve never watched it live on TV, not was I in the US at the time when it aired. I’ve only heard of it thanks to the idiom it lead to.", "votes": 1 } ] }, { "id": 323363, "postDate": "2018-05-04T23:41:34.103Z", "content": "|Sarcasm| Very Disappointed |Sarcasm|
\nFixed your title
\n\n\nFixed your title
\n
You win the daily internet police award.
\n\nMay the 4th be with you!
\n\nPS Hopefully you don't get tempted to fix the spelling in the above sentence.
", "rawMarkdown": "> Fixed your title\n\nYou win the daily internet police award.\n\nMay the 4th be with you!\n\nPS Hopefully you don't get tempted to fix the spelling in the above sentence.", "votes": 4 }, { "id": 323368, "postDate": "2018-05-05T00:01:42.067Z", "content": "Thanks pocket. I've always maintained that someone needs to establish the official sarcasm font.
", "rawMarkdown": "Thanks pocket. I've always maintained that someone needs to establish the official sarcasm font. ", "votes": 3 }, { "id": 323541, "postDate": "2018-05-05T12:37:35.067Z", "content": "That's typical Arsenal...
", "rawMarkdown": "That's typical Arsenal..." } ] }, { "id": 323725, "postDate": "2018-05-06T02:17:10.523Z", "content": "ip%27 ??? seriously
", "rawMarkdown": "ip%27 ??? seriously", "replies": [ { "id": 324001, "postDate": "2018-05-06T22:17:48.797Z", "rawMarkdown": "", "votes": -9, "isDeleted": true } ] } ], "comments": [ { "id": 323287, "author_name": "inversion", "author_url": "", "post_date": "2018-05-04T18:57:17.160000", "content": "@Bojan - We'll do our best to make it up to you in future competitions. ;-)
", "votes": 14, "replies": [ { "id": 323296, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-04T19:21:12.713000", "content": "Hope so. Would hate to see this become a trend.
", "votes": 7, "replies": [] } ] }, { "id": 323512, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2018-05-05T10:17:46.367000", "content": "You mean \"except IP%27 leak\"?
", "votes": 8, "replies": [ { "id": 323528, "author_name": "CPMP", "author_url": "", "post_date": "2018-05-05T11:22:11.187000", "content": "shhhhh, don't tell anyone!
", "votes": 4, "replies": [] }, { "id": 323549, "author_name": "Akarsh Raj", "author_url": "", "post_date": "2018-05-05T13:04:18.570000", "content": "Not again xD
", "votes": 3, "replies": [] }, { "id": 323724, "author_name": "Peter", "author_url": "", "post_date": "2018-05-06T02:16:46.817000", "content": "ip%27 leak??? serious?
", "votes": 1, "replies": [] }, { "id": 323753, "author_name": "Ahmet Erdem", "author_url": "", "post_date": "2018-05-06T05:07:13.770000", "content": "@Peter it was a joke from another post
", "votes": 2, "replies": [] } ] }, { "id": 325514, "author_name": "CPMP", "author_url": "", "post_date": "2018-05-08T13:43:52.953000", "content": "Sorry to really disappoint you Bojan, but there was a (minor) leak: test data was sorted by click time AND is_attributed. Exploiting this could yield some boost, but nothing like previous major leaks we saw. In my case the boost was about 0.0004.
\n\nThe leak was publicly shared by @plantsgo soon enough for people to take advantage of it.
", "votes": 5, "replies": [ { "id": 326322, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-09T15:10:47.743000", "content": "Well played, very well played. I really should pay more attention to the discussion next time. There were several small tips and clues that could have made the difference. The problem with this competition was that the dataset was so large, it was not easy to iterate quickly and try many things, even the ones I was aware of.
", "votes": 2, "replies": [] }, { "id": 326336, "author_name": "CPMP", "author_url": "", "post_date": "2018-05-09T15:25:02.183000", "content": "I have been thinking of it for a month, see here, However, @plantsgo share made me try the one thing I was looking at, namely modifying the target in train.
\n\nThat said, I don't know if I would have tried it if he had not shared. As you say, there were so many things to try, and so little time, that prioritizing was difficult.
", "votes": 2, "replies": [] } ] }, { "id": 323290, "author_name": "Tilii", "author_url": "", "post_date": "2018-05-04T19:02:38.107000", "content": "
This is a classic reference :-) Only a few upvotes Tilii, I wonder how many of many of us are old enough (and North American) to have watched it on TV?
", "votes": 2, "replies": [] }, { "id": 323716, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-06T00:51:57.463000", "content": "That actually is Tilii in the gif. ;-)
\n\nI’ve never watched it live on TV, not was I in the US at the time when it aired. I’ve only heard of it thanks to the idiom it lead to.
", "votes": 1, "replies": [] } ] }, { "id": 323363, "author_name": "pocket", "author_url": "", "post_date": "2018-05-04T23:41:34.103000", "content": "|Sarcasm| Very Disappointed |Sarcasm|
\nFixed your title
\n\n\nFixed your title
\n
You win the daily internet police award.
\n\nMay the 4th be with you!
\n\nPS Hopefully you don't get tempted to fix the spelling in the above sentence.
", "votes": 4, "replies": [] }, { "id": 323368, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-05T00:01:42.067000", "content": "Thanks pocket. I've always maintained that someone needs to establish the official sarcasm font.
", "votes": 3, "replies": [] }, { "id": 323541, "author_name": "Laevatein", "author_url": "", "post_date": "2018-05-05T12:37:35.067000", "content": "That's typical Arsenal...
", "votes": 0, "replies": [] } ] }, { "id": 323725, "author_name": "Peter", "author_url": "", "post_date": "2018-05-06T02:17:10.523000", "content": "ip%27 ??? seriously
", "votes": 0, "replies": [ { "id": 324001, "author_name": "", "author_url": "", "post_date": "2018-05-06T22:17:48.797000", "content": "", "votes": -9, "replies": [] } ] } ], "raw_markdown_by_id": { "323281": "Here we are, just a few days before the end, and it seems that there have not been any leaks in this competition. The way things stand right now, it looks like the team that has the best model and has gleaned the most from the data, will, in fact, win this thing. The organizer and Kaggle have done an impeccable job of preparing this dataset. This goes against all the expectations that we have from TalkingData competitions. ", "323287": "@Bojan - We'll do our best to make it up to you in future competitions. ;-)", "323512": "You mean \"except IP%27 leak\"?", "325514": "Sorry to really disappoint you Bojan, but there was a (minor) leak: test data was sorted by click time AND is_attributed. Exploiting this could yield some boost, but nothing like previous major leaks we saw. In my case the boost was about 0.0004.\n\nThe leak was publicly shared by @plantsgo soon enough for people to take advantage of it.", "323290": "\n![enter image description here][1]\n\n\n\n [1]: https://media.giphy.com/media/Xxei1HDpBhMys/giphy.gif \"Jumping the shark\"", "323363": "|Sarcasm| Very Disappointed |Sarcasm| \nFixed your title", "323725": "ip%27 ??? seriously" } }