{
  "id": 56237,
  "title": "Happy and exhausted",
  "url": "/competitions/talkingdata-adtracking-fraud-detection/discussion/56237",
  "author_name": "CPMP",
  "post_date": "2018-05-08T00:06:18.810000",
  "votes": 38,
  "comment_count": 40,
  "views": 0,
  "content": "<p>Congrats to the top teams, and also to all who managed to work around this huge dataset.</p>\n\n<p>What I think would be my best sub failed to upload, but it would not be good enough to change my rank.  I'll submit it still to get an idea, but no regret.</p>\n\n<p>I will share my solution in the coming days.  I look froward to see other teams solutions!</p>\n\n<p>Thanks to the organizers and the Kaggle team for an almost flawless (if it hadn't been for Dirk share).  Thanks also to all who shared (not at the last inute), I got many ideas from you!</p>\n\n<p>Edit: My last sub was submitted before deadline, and scoring started shortly after.  It scores 0.9829 public, and 0.9837 private.  As I said above, it would not change my rank.</p>\n\n<p>Edit: I <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/56283#latest-325289\">shared my solution</a>.</p>",
  "messages": [
    {
      "id": 324792,
      "postDate": "2018-05-08T00:06:18.810Z",
      "content": "<p>Congrats to the top teams, and also to all who managed to work around this huge dataset.</p>\n\n<p>What I think would be my best sub failed to upload, but it would not be good enough to change my rank.  I'll submit it still to get an idea, but no regret.</p>\n\n<p>I will share my solution in the coming days.  I look froward to see other teams solutions!</p>\n\n<p>Thanks to the organizers and the Kaggle team for an almost flawless (if it hadn't been for Dirk share).  Thanks also to all who shared (not at the last inute), I got many ideas from you!</p>\n\n<p>Edit: My last sub was submitted before deadline, and scoring started shortly after.  It scores 0.9829 public, and 0.9837 private.  As I said above, it would not change my rank.</p>\n\n<p>Edit: I <a href=\"https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/56283#latest-325289\">shared my solution</a>.</p>",
      "rawMarkdown": "Congrats to the top teams, and also to all who managed to work around this huge dataset.\n\nWhat I think would be my best sub failed to upload, but it would not be good enough to change my rank.  I'll submit it still to get an idea, but no regret.\n\nI will share my solution in the coming days.  I look froward to see other teams solutions!\n\nThanks to the organizers and the Kaggle team for an almost flawless (if it hadn't been for Dirk share).  Thanks also to all who shared (not at the last inute), I got many ideas from you!\n\nEdit: My last sub was submitted before deadline, and scoring started shortly after.  It scores 0.9829 public, and 0.9837 private.  As I said above, it would not change my rank.\n\nEdit: I [shared my solution][1].\n\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/56283#latest-325289",
      "votes": 38
    },
    {
      "id": 324815,
      "postDate": "2018-05-08T00:18:27.687Z",
      "content": "<p>what an ending! we manage to squeeze in two submissions in the last 20 minutes which turned out to be our best submissions - we suffered a drop but hey we only spent 10 days on this  :)</p>",
      "rawMarkdown": "what an ending! we manage to squeeze in two submissions in the last 20 minutes which turned out to be our best submissions - we suffered a drop but hey we only spent 10 days on this  :)",
      "votes": 5,
      "replies": [
        {
          "id": 324847,
          "postDate": "2018-05-08T00:39:57.150Z",
          "content": "<p>Thanks.  Never give up till the end!  Glad it worked for you.</p>",
          "rawMarkdown": "Thanks.  Never give up till the end!  Glad it worked for you."
        }
      ]
    },
    {
      "id": 324807,
      "postDate": "2018-05-08T00:14:55.917Z",
      "content": "<p>Congrats! Really glad for you. </p>",
      "rawMarkdown": "Congrats! Really glad for you. ",
      "votes": 3,
      "replies": [
        {
          "id": 324838,
          "postDate": "2018-05-08T00:35:43.580Z",
          "content": "<p>Thanks!  Sorry I could not provide the kernel you asked for today ;)</p>",
          "rawMarkdown": "Thanks!  Sorry I could not provide the kernel you asked for today ;)",
          "votes": 1
        },
        {
          "id": 325294,
          "postDate": "2018-05-08T09:28:37.253Z",
          "content": "<p>I see you're one away from gold.  I won't say I hope for you that someone has cheated above you, but I write it ;)</p>",
          "rawMarkdown": "I see you're one away from gold.  I won't say I hope for you that someone has cheated above you, but I write it ;)",
          "votes": 2
        },
        {
          "id": 326321,
          "postDate": "2018-05-09T15:08:58.413Z",
          "content": "<blockquote>\n  <p><strong>CPMP wrote</strong></p>\n  \n  <blockquote>\n    <p>Thanks!  Sorry I could not provide the kernel you asked for today ;)</p>\n  </blockquote>\n</blockquote>\n\n<p>I'll make sure that you make up for it in some later competition. ;-)</p>\n\n<p>Alas, all the teams ahead of us seem to have been hones, so no gold for us. :-)</p>",
          "rawMarkdown": "\n&gt; **CPMP wrote**\n&gt; \n&gt; &gt; Thanks!  Sorry I could not provide the kernel you asked for today ;)\n\nI'll make sure that you make up for it in some later competition. ;-)\n\nAlas, all the teams ahead of us seem to have been hones, so no gold for us. :-)",
          "votes": 2
        }
      ]
    },
    {
      "id": 324879,
      "postDate": "2018-05-08T01:08:09.777Z",
      "content": "<p>Congrats~ </p>\n\n<p>The best response to those meaningless doubt</p>",
      "rawMarkdown": "Congrats~ \n\nThe best response to those meaningless doubt",
      "votes": 4,
      "replies": [
        {
          "id": 324890,
          "postDate": "2018-05-08T01:20:40.827Z",
          "content": "<p>;)</p>",
          "rawMarkdown": ";)"
        }
      ]
    },
    {
      "id": 325165,
      "postDate": "2018-05-08T07:23:23.573Z",
      "content": "<p>Congratulations! </p>\n\n<p>We have weird experience with CV score during this competition.\nLooking forward to see your solution!</p>\n\n<p>Cheers ;)</p>",
      "rawMarkdown": "Congratulations! \n\nWe have weird experience with CV score during this competition.\nLooking forward to see your solution!\n\nCheers ;)",
      "votes": 1
    },
    {
      "id": 325121,
      "postDate": "2018-05-08T06:27:16.187Z",
      "content": "<p>Congrats :). <br> My CV did not work for me this time. Looking forward for your solution.</p>",
      "rawMarkdown": "Congrats :). <br> My CV did not work for me this time. Looking forward for your solution.",
      "votes": 1,
      "replies": [
        {
          "id": 325457,
          "postDate": "2018-05-08T12:42:26.333Z",
          "content": "<p>What happened ? as far as i remember , you have provided many good inputs for this competition especially Adversarial Validation .. </p>",
          "rawMarkdown": "What happened ? as far as i remember , you have provided many good inputs for this competition especially Adversarial Validation .. "
        }
      ]
    },
    {
      "id": 325075,
      "postDate": "2018-05-08T05:08:31.690Z",
      "content": "<p>Congrats, waiting for your solution impatiently​.</p>",
      "rawMarkdown": "Congrats, waiting for your solution impatiently​.",
      "votes": 1
    },
    {
      "id": 325042,
      "postDate": "2018-05-08T03:53:01.607Z",
      "content": "<p>Congrats CPMP and I want to say BIG thank you for all your shared ideas &amp; valuable insights through all these past competitions!</p>\n\n<p>To be honest, your presence and other top gold Kagglers are the ones that make Kaggle's competition is still interesting to compete on, because nowadays everyone can easily get silver &amp; bronze just by blending other kernels or csvs without even thinking.</p>",
      "rawMarkdown": "Congrats CPMP and I want to say BIG thank you for all your shared ideas &amp; valuable insights through all these past competitions!\n\nTo be honest, your presence and other top gold Kagglers are the ones that make Kaggle's competition is still interesting to compete on, because nowadays everyone can easily get silver &amp; bronze just by blending other kernels or csvs without even thinking.",
      "votes": 1
    },
    {
      "id": 325039,
      "postDate": "2018-05-08T03:46:18.650Z",
      "content": "<p>Congrats @CPMP !!!! Looking forward to hear your tips and advice's in the next competition....</p>",
      "rawMarkdown": "Congrats @CPMP !!!! Looking forward to hear your tips and advice's in the next competition....",
      "votes": 1
    },
    {
      "id": 324896,
      "postDate": "2018-05-08T01:23:16.040Z",
      "content": "<p>great finish, CPMP! Always amazed to read your experience.</p>",
      "rawMarkdown": "great finish, CPMP! Always amazed to read your experience.",
      "votes": 1
    },
    {
      "id": 324873,
      "postDate": "2018-05-08T01:00:00.410Z",
      "content": "<p>thanks and waiting for your solution:) And the advice is very gooooood =]</p>",
      "rawMarkdown": "thanks and waiting for your solution:) And the advice is very gooooood =]",
      "votes": 1
    },
    {
      "id": 324840,
      "postDate": "2018-05-08T00:36:20.517Z",
      "content": "<p>Thanks for your advice and your contribution to the community.</p>",
      "rawMarkdown": "Thanks for your advice and your contribution to the community.",
      "votes": 1
    },
    {
      "id": 324834,
      "postDate": "2018-05-08T00:34:17.007Z",
      "content": "<p>@CPMP congrats for such a strong finish. Well deserved.</p>",
      "rawMarkdown": "@CPMP congrats for such a strong finish. Well deserved.",
      "votes": 1
    },
    {
      "id": 324818,
      "postDate": "2018-05-08T00:19:22.423Z",
      "content": "<p>Congrats! Well deserved the solo gold medal, and thank you for continuous contribution to discussions. Looking forward to your post later. </p>",
      "rawMarkdown": "Congrats! Well deserved the solo gold medal, and thank you for continuous contribution to discussions. Looking forward to your post later. ",
      "votes": 1
    },
    {
      "id": 324809,
      "postDate": "2018-05-08T00:16:21.370Z",
      "content": "<p>CPMP, I learned a lot from you in this competition. Thank you for your help!</p>",
      "rawMarkdown": "CPMP, I learned a lot from you in this competition. Thank you for your help!",
      "votes": 1,
      "replies": [
        {
          "id": 324839,
          "postDate": "2018-05-08T00:36:07.667Z",
          "content": "<p>Thanks!  Many learned form you too!</p>",
          "rawMarkdown": "Thanks!  Many learned form you too!"
        },
        {
          "id": 324849,
          "postDate": "2018-05-08T00:41:52.503Z",
          "content": "<p>Hope to compete with you in future competitions~ And may I ask do you use cross validation or time-wise validation(day7,8 train and day 9 validation)?</p>",
          "rawMarkdown": "Hope to compete with you in future competitions~ And may I ask do you use cross validation or time-wise validation(day7,8 train and day 9 validation)?"
        },
        {
          "id": 324860,
          "postDate": "2018-05-08T00:47:12.897Z",
          "content": "<p>I'll share my solution, for now I'm enjoying ;)</p>",
          "rawMarkdown": "I'll share my solution, for now I'm enjoying ;)",
          "votes": 3
        }
      ]
    },
    {
      "id": 324806,
      "postDate": "2018-05-08T00:14:00.373Z",
      "content": "<p>Congrats and thank you, CPMP. Learnt a lot from you guys.</p>",
      "rawMarkdown": "Congrats and thank you, CPMP. Learnt a lot from you guys.",
      "votes": 1,
      "replies": [
        {
          "id": 324836,
          "postDate": "2018-05-08T00:34:57.447Z",
          "content": "<p>Thanks!</p>",
          "rawMarkdown": "Thanks!"
        }
      ]
    },
    {
      "id": 325032,
      "postDate": "2018-05-08T03:38:09.767Z",
      "content": "<p>i am sure , your kernel and solutions will help us realize the mistakes we made, and be a guiding light for the steps to follow in the future competitions. greatful for your contributions in the discussions section! Ending this competition with disapppointment as i fell from the 7% mark, but also with extreme curiosity as i look forward to the solutions from the masters. Thank you.</p>",
      "rawMarkdown": "i am sure , your kernel and solutions will help us realize the mistakes we made, and be a guiding light for the steps to follow in the future competitions. greatful for your contributions in the discussions section! Ending this competition with disapppointment as i fell from the 7% mark, but also with extreme curiosity as i look forward to the solutions from the masters. Thank you.",
      "votes": 2
    },
    {
      "id": 324875,
      "postDate": "2018-05-08T01:00:45.413Z",
      "content": "<p>Congrats, amazing result ! Would be interested to see what kind of magic features you found ;)</p>\n\n<p>Also if you don't mind explaining us your logic for the CV set choice ? I'd like to understand what would be a correct procedure. In my case I trained on all data with CV set on day 9, hour 4 / 9 / 13, which seemed to be half of the hours present in the public+private leaderboard. Then I would get the number of rounds and retrain on the whole data set with no CV set but limited number of rounds? Does that makes sense ?</p>",
      "rawMarkdown": "Congrats, amazing result ! Would be interested to see what kind of magic features you found ;)\n\nAlso if you don't mind explaining us your logic for the CV set choice ? I'd like to understand what would be a correct procedure. In my case I trained on all data with CV set on day 9, hour 4 / 9 / 13, which seemed to be half of the hours present in the public+private leaderboard. Then I would get the number of rounds and retrain on the whole data set with no CV set but limited number of rounds? Does that makes sense ?",
      "votes": 2
    },
    {
      "id": 324812,
      "postDate": "2018-05-08T00:17:54.567Z",
      "content": "<p>Congrats CPMP  well deserved ...Solo working in this huge dataset and achieving such great performance is quit a big challenge </p>",
      "rawMarkdown": "Congrats CPMP  well deserved ...Solo working in this huge dataset and achieving such great performance is quit a big challenge ",
      "votes": 2,
      "replies": [
        {
          "id": 324844,
          "postDate": "2018-05-08T00:38:48.680Z",
          "content": "<p>Thanks!</p>",
          "rawMarkdown": "Thanks!"
        }
      ]
    },
    {
      "id": 324805,
      "postDate": "2018-05-08T00:13:16.260Z",
      "content": "<p>Congrats with your solo Gold ) Hard work to make gold here in solo mode, I know )</p>",
      "rawMarkdown": "Congrats with your solo Gold ) Hard work to make gold here in solo mode, I know )",
      "votes": 2,
      "replies": [
        {
          "id": 324835,
          "postDate": "2018-05-08T00:34:43.303Z",
          "content": "<p>Your team pushed me ;)  Congrats on your result, you teamed, but you climbed very fast in the last few days.</p>",
          "rawMarkdown": "Your team pushed me ;)  Congrats on your result, you teamed, but you climbed very fast in the last few days.",
          "votes": 1
        }
      ]
    },
    {
      "id": 324795,
      "postDate": "2018-05-08T00:08:50.357Z",
      "content": "<p>Congrats on your Gold CPMP .  You have been a great mentor here . i suggest you make the newbie section more elaborative . so that it becomes the guide for everyone . Not in the top team though but We managed to finish silver after the rollercoaster in the last 12 hours. Our best submission was the one we did not mark , nonetheless i am happy.\nThank You.</p>",
      "rawMarkdown": "Congrats on your Gold CPMP .  You have been a great mentor here . i suggest you make the newbie section more elaborative . so that it becomes the guide for everyone . Not in the top team though but We managed to finish silver after the rollercoaster in the last 12 hours. Our best submission was the one we did not mark , nonetheless i am happy.\nThank You.",
      "votes": 2,
      "replies": [
        {
          "id": 324802,
          "postDate": "2018-05-08T00:11:16.603Z",
          "content": "<p>Thanks.  Well done, this was a challenging competition!</p>",
          "rawMarkdown": "Thanks.  Well done, this was a challenging competition!",
          "votes": 1
        }
      ]
    },
    {
      "id": 324810,
      "postDate": "2018-05-08T00:16:39.907Z",
      "content": "<p>Congrats! Hopefully it'll be my first medal (rip for my removed team of DSB2018). I hope there's not a <code>2nd stage model submission</code> anymore :s <br>\nAnyway, this competition burns me ~100$ credit for high RAM VM instance on tencent cloud, not to mention my 300$ for GPU instance's of DSB2018. today's competitions are expensive lol</p>",
      "rawMarkdown": "Congrats! Hopefully it'll be my first medal (rip for my removed team of DSB2018). I hope there's not a `2nd stage model submission` anymore :s  \nAnyway, this competition burns me ~100$ credit for high RAM VM instance on tencent cloud, not to mention my 300$ for GPU instance's of DSB2018. today's competitions are expensive lol",
      "replies": [
        {
          "id": 324842,
          "postDate": "2018-05-08T00:36:43.447Z",
          "content": "<p>Thanks.  No second stage story here, don't worry.</p>",
          "rawMarkdown": "Thanks.  No second stage story here, don't worry."
        },
        {
          "id": 324859,
          "postDate": "2018-05-08T00:47:02.197Z",
          "content": "<p>Same here, $100 credit at google cloud. It turns out to be worthy. Feel happy to defeat those last minute's submissions.</p>",
          "rawMarkdown": "Same here, $100 credit at google cloud. It turns out to be worthy. Feel happy to defeat those last minute's submissions."
        },
        {
          "id": 325033,
          "postDate": "2018-05-08T03:42:34.817Z",
          "content": "<p>I learned a lot from Heng's maskrcnn and your team's UNet in the last competition, waiting for your post on this one, hope to learn more ;)</p>",
          "rawMarkdown": "I learned a lot from Heng's maskrcnn and your team's UNet in the last competition, waiting for your post on this one, hope to learn more ;)"
        }
      ]
    },
    {
      "id": 324888,
      "postDate": "2018-05-08T01:14:45.583Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    },
    {
      "id": 324881,
      "postDate": "2018-05-08T01:09:55.927Z",
      "rawMarkdown": "",
      "votes": 1,
      "isDeleted": true
    },
    {
      "id": 325390,
      "postDate": "2018-05-08T11:13:40.667Z",
      "content": "<p>Thanks for sharing.\nCongrats !</p>",
      "rawMarkdown": "Thanks for sharing.\nCongrats !",
      "votes": 1
    }
  ],
  "comments": [
    {
      "id": 324815,
      "author_name": "Yifan Xie",
      "author_url": "",
      "post_date": "2018-05-08T00:18:27.687000",
      "content": "<p>what an ending! we manage to squeeze in two submissions in the last 20 minutes which turned out to be our best submissions - we suffered a drop but hey we only spent 10 days on this  :)</p>",
      "votes": 5,
      "replies": [
        {
          "id": 324847,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:39:57.150000",
          "content": "<p>Thanks.  Never give up till the end!  Glad it worked for you.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 324807,
      "author_name": "Bojan Tunguz",
      "author_url": "",
      "post_date": "2018-05-08T00:14:55.917000",
      "content": "<p>Congrats! Really glad for you. </p>",
      "votes": 3,
      "replies": [
        {
          "id": 324838,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:35:43.580000",
          "content": "<p>Thanks!  Sorry I could not provide the kernel you asked for today ;)</p>",
          "votes": 1,
          "replies": []
        },
        {
          "id": 325294,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T09:28:37.253000",
          "content": "<p>I see you're one away from gold.  I won't say I hope for you that someone has cheated above you, but I write it ;)</p>",
          "votes": 2,
          "replies": []
        },
        {
          "id": 326321,
          "author_name": "Bojan Tunguz",
          "author_url": "",
          "post_date": "2018-05-09T15:08:58.413000",
          "content": "<blockquote>\n  <p><strong>CPMP wrote</strong></p>\n  \n  <blockquote>\n    <p>Thanks!  Sorry I could not provide the kernel you asked for today ;)</p>\n  </blockquote>\n</blockquote>\n\n<p>I'll make sure that you make up for it in some later competition. ;-)</p>\n\n<p>Alas, all the teams ahead of us seem to have been hones, so no gold for us. :-)</p>",
          "votes": 2,
          "replies": []
        }
      ]
    },
    {
      "id": 324879,
      "author_name": "Hikkiiiiiiiii",
      "author_url": "",
      "post_date": "2018-05-08T01:08:09.777000",
      "content": "<p>Congrats~ </p>\n\n<p>The best response to those meaningless doubt</p>",
      "votes": 4,
      "replies": [
        {
          "id": 324890,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T01:20:40.827000",
          "content": "<p>;)</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 325165,
      "author_name": "Meyk",
      "author_url": "",
      "post_date": "2018-05-08T07:23:23.573000",
      "content": "<p>Congratulations! </p>\n\n<p>We have weird experience with CV score during this competition.\nLooking forward to see your solution!</p>\n\n<p>Cheers ;)</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 325121,
      "author_name": "Sohaib Omar",
      "author_url": "",
      "post_date": "2018-05-08T06:27:16.187000",
      "content": "<p>Congrats :). <br> My CV did not work for me this time. Looking forward for your solution.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 325457,
          "author_name": "Mayank Soni",
          "author_url": "",
          "post_date": "2018-05-08T12:42:26.333000",
          "content": "<p>What happened ? as far as i remember , you have provided many good inputs for this competition especially Adversarial Validation .. </p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 325075,
      "author_name": "Araks Stepanyan",
      "author_url": "",
      "post_date": "2018-05-08T05:08:31.690000",
      "content": "<p>Congrats, waiting for your solution impatiently​.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 325042,
      "author_name": "SDA",
      "author_url": "",
      "post_date": "2018-05-08T03:53:01.607000",
      "content": "<p>Congrats CPMP and I want to say BIG thank you for all your shared ideas &amp; valuable insights through all these past competitions!</p>\n\n<p>To be honest, your presence and other top gold Kagglers are the ones that make Kaggle's competition is still interesting to compete on, because nowadays everyone can easily get silver &amp; bronze just by blending other kernels or csvs without even thinking.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 325039,
      "author_name": "Samrat Pandiri",
      "author_url": "",
      "post_date": "2018-05-08T03:46:18.650000",
      "content": "<p>Congrats @CPMP !!!! Looking forward to hear your tips and advice's in the next competition....</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324896,
      "author_name": "DSEverything",
      "author_url": "",
      "post_date": "2018-05-08T01:23:16.040000",
      "content": "<p>great finish, CPMP! Always amazed to read your experience.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324873,
      "author_name": "Perfect Is Shit",
      "author_url": "",
      "post_date": "2018-05-08T01:00:00.410000",
      "content": "<p>thanks and waiting for your solution:) And the advice is very gooooood =]</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324840,
      "author_name": "seaguII",
      "author_url": "",
      "post_date": "2018-05-08T00:36:20.517000",
      "content": "<p>Thanks for your advice and your contribution to the community.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324834,
      "author_name": "YaGana Sheriff-Hussaini",
      "author_url": "",
      "post_date": "2018-05-08T00:34:17.007000",
      "content": "<p>@CPMP congrats for such a strong finish. Well deserved.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324818,
      "author_name": "Rui Li",
      "author_url": "",
      "post_date": "2018-05-08T00:19:22.423000",
      "content": "<p>Congrats! Well deserved the solo gold medal, and thank you for continuous contribution to discussions. Looking forward to your post later. </p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324809,
      "author_name": "Snorlax",
      "author_url": "",
      "post_date": "2018-05-08T00:16:21.370000",
      "content": "<p>CPMP, I learned a lot from you in this competition. Thank you for your help!</p>",
      "votes": 1,
      "replies": [
        {
          "id": 324839,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:36:07.667000",
          "content": "<p>Thanks!  Many learned form you too!</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 324849,
          "author_name": "Snorlax",
          "author_url": "",
          "post_date": "2018-05-08T00:41:52.503000",
          "content": "<p>Hope to compete with you in future competitions~ And may I ask do you use cross validation or time-wise validation(day7,8 train and day 9 validation)?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 324860,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:47:12.897000",
          "content": "<p>I'll share my solution, for now I'm enjoying ;)</p>",
          "votes": 3,
          "replies": []
        }
      ]
    },
    {
      "id": 324806,
      "author_name": "Laevatein",
      "author_url": "",
      "post_date": "2018-05-08T00:14:00.373000",
      "content": "<p>Congrats and thank you, CPMP. Learnt a lot from you guys.</p>",
      "votes": 1,
      "replies": [
        {
          "id": 324836,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:34:57.447000",
          "content": "<p>Thanks!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 325032,
      "author_name": "nickhillator",
      "author_url": "",
      "post_date": "2018-05-08T03:38:09.767000",
      "content": "<p>i am sure , your kernel and solutions will help us realize the mistakes we made, and be a guiding light for the steps to follow in the future competitions. greatful for your contributions in the discussions section! Ending this competition with disapppointment as i fell from the 7% mark, but also with extreme curiosity as i look forward to the solutions from the masters. Thank you.</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 324875,
      "author_name": "Antoine",
      "author_url": "",
      "post_date": "2018-05-08T01:00:45.413000",
      "content": "<p>Congrats, amazing result ! Would be interested to see what kind of magic features you found ;)</p>\n\n<p>Also if you don't mind explaining us your logic for the CV set choice ? I'd like to understand what would be a correct procedure. In my case I trained on all data with CV set on day 9, hour 4 / 9 / 13, which seemed to be half of the hours present in the public+private leaderboard. Then I would get the number of rounds and retrain on the whole data set with no CV set but limited number of rounds? Does that makes sense ?</p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 324812,
      "author_name": "Serigne ",
      "author_url": "",
      "post_date": "2018-05-08T00:17:54.567000",
      "content": "<p>Congrats CPMP  well deserved ...Solo working in this huge dataset and achieving such great performance is quit a big challenge </p>",
      "votes": 2,
      "replies": [
        {
          "id": 324844,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:38:48.680000",
          "content": "<p>Thanks!</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 324805,
      "author_name": "Kruegger",
      "author_url": "",
      "post_date": "2018-05-08T00:13:16.260000",
      "content": "<p>Congrats with your solo Gold ) Hard work to make gold here in solo mode, I know )</p>",
      "votes": 2,
      "replies": [
        {
          "id": 324835,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:34:43.303000",
          "content": "<p>Your team pushed me ;)  Congrats on your result, you teamed, but you climbed very fast in the last few days.</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 324795,
      "author_name": "Mayank Soni",
      "author_url": "",
      "post_date": "2018-05-08T00:08:50.357000",
      "content": "<p>Congrats on your Gold CPMP .  You have been a great mentor here . i suggest you make the newbie section more elaborative . so that it becomes the guide for everyone . Not in the top team though but We managed to finish silver after the rollercoaster in the last 12 hours. Our best submission was the one we did not mark , nonetheless i am happy.\nThank You.</p>",
      "votes": 2,
      "replies": [
        {
          "id": 324802,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:11:16.603000",
          "content": "<p>Thanks.  Well done, this was a challenging competition!</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 324810,
      "author_name": "xuan",
      "author_url": "",
      "post_date": "2018-05-08T00:16:39.907000",
      "content": "<p>Congrats! Hopefully it'll be my first medal (rip for my removed team of DSB2018). I hope there's not a <code>2nd stage model submission</code> anymore :s <br>\nAnyway, this competition burns me ~100$ credit for high RAM VM instance on tencent cloud, not to mention my 300$ for GPU instance's of DSB2018. today's competitions are expensive lol</p>",
      "votes": 0,
      "replies": [
        {
          "id": 324842,
          "author_name": "CPMP",
          "author_url": "",
          "post_date": "2018-05-08T00:36:43.447000",
          "content": "<p>Thanks.  No second stage story here, don't worry.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 324859,
          "author_name": "seaguII",
          "author_url": "",
          "post_date": "2018-05-08T00:47:02.197000",
          "content": "<p>Same here, $100 credit at google cloud. It turns out to be worthy. Feel happy to defeat those last minute's submissions.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 325033,
          "author_name": "xuan",
          "author_url": "",
          "post_date": "2018-05-08T03:42:34.817000",
          "content": "<p>I learned a lot from Heng's maskrcnn and your team's UNet in the last competition, waiting for your post on this one, hope to learn more ;)</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 324888,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-05-08T01:14:45.583000",
      "content": "",
      "votes": 1,
      "replies": []
    },
    {
      "id": 324881,
      "author_name": "",
      "author_url": "",
      "post_date": "2018-05-08T01:09:55.927000",
      "content": "",
      "votes": 1,
      "replies": []
    },
    {
      "id": 325390,
      "author_name": "Bruno16",
      "author_url": "",
      "post_date": "2018-05-08T11:13:40.667000",
      "content": "<p>Thanks for sharing.\nCongrats !</p>",
      "votes": 1,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "324792": "Congrats to the top teams, and also to all who managed to work around this huge dataset.\n\nWhat I think would be my best sub failed to upload, but it would not be good enough to change my rank.  I'll submit it still to get an idea, but no regret.\n\nI will share my solution in the coming days.  I look froward to see other teams solutions!\n\nThanks to the organizers and the Kaggle team for an almost flawless (if it hadn't been for Dirk share).  Thanks also to all who shared (not at the last inute), I got many ideas from you!\n\nEdit: My last sub was submitted before deadline, and scoring started shortly after.  It scores 0.9829 public, and 0.9837 private.  As I said above, it would not change my rank.\n\nEdit: I [shared my solution][1].\n\n\n  [1]: https://www.kaggle.com/c/talkingdata-adtracking-fraud-detection/discussion/56283#latest-325289",
    "324815": "what an ending! we manage to squeeze in two submissions in the last 20 minutes which turned out to be our best submissions - we suffered a drop but hey we only spent 10 days on this  :)",
    "324807": "Congrats! Really glad for you. ",
    "324879": "Congrats~ \n\nThe best response to those meaningless doubt",
    "325165": "Congratulations! \n\nWe have weird experience with CV score during this competition.\nLooking forward to see your solution!\n\nCheers ;)",
    "325121": "Congrats :). <br> My CV did not work for me this time. Looking forward for your solution.",
    "325075": "Congrats, waiting for your solution impatiently​.",
    "325042": "Congrats CPMP and I want to say BIG thank you for all your shared ideas &amp; valuable insights through all these past competitions!\n\nTo be honest, your presence and other top gold Kagglers are the ones that make Kaggle's competition is still interesting to compete on, because nowadays everyone can easily get silver &amp; bronze just by blending other kernels or csvs without even thinking.",
    "325039": "Congrats @CPMP !!!! Looking forward to hear your tips and advice's in the next competition....",
    "324896": "great finish, CPMP! Always amazed to read your experience.",
    "324873": "thanks and waiting for your solution:) And the advice is very gooooood =]",
    "324840": "Thanks for your advice and your contribution to the community.",
    "324834": "@CPMP congrats for such a strong finish. Well deserved.",
    "324818": "Congrats! Well deserved the solo gold medal, and thank you for continuous contribution to discussions. Looking forward to your post later. ",
    "324809": "CPMP, I learned a lot from you in this competition. Thank you for your help!",
    "324806": "Congrats and thank you, CPMP. Learnt a lot from you guys.",
    "325032": "i am sure , your kernel and solutions will help us realize the mistakes we made, and be a guiding light for the steps to follow in the future competitions. greatful for your contributions in the discussions section! Ending this competition with disapppointment as i fell from the 7% mark, but also with extreme curiosity as i look forward to the solutions from the masters. Thank you.",
    "324875": "Congrats, amazing result ! Would be interested to see what kind of magic features you found ;)\n\nAlso if you don't mind explaining us your logic for the CV set choice ? I'd like to understand what would be a correct procedure. In my case I trained on all data with CV set on day 9, hour 4 / 9 / 13, which seemed to be half of the hours present in the public+private leaderboard. Then I would get the number of rounds and retrain on the whole data set with no CV set but limited number of rounds? Does that makes sense ?",
    "324812": "Congrats CPMP  well deserved ...Solo working in this huge dataset and achieving such great performance is quit a big challenge ",
    "324805": "Congrats with your solo Gold ) Hard work to make gold here in solo mode, I know )",
    "324795": "Congrats on your Gold CPMP .  You have been a great mentor here . i suggest you make the newbie section more elaborative . so that it becomes the guide for everyone . Not in the top team though but We managed to finish silver after the rollercoaster in the last 12 hours. Our best submission was the one we did not mark , nonetheless i am happy.\nThank You.",
    "324810": "Congrats! Hopefully it'll be my first medal (rip for my removed team of DSB2018). I hope there's not a `2nd stage model submission` anymore :s  \nAnyway, this competition burns me ~100$ credit for high RAM VM instance on tencent cloud, not to mention my 300$ for GPU instance's of DSB2018. today's competitions are expensive lol",
    "324888": "",
    "324881": "",
    "325390": "Thanks for sharing.\nCongrats !"
  }
}