Congrats to all, esp the ones with amazing top scores, above or near 0.9. Let me aslo thank the organizers, CERN and Kaggle team for hosting such a challenge. I also want to thank those who shared so much here, including Yuval, Nicole Finnie, Heng Cher Keng, The Chemist (aka Grzegorz Sionkowski), Outrunner, Luis Andre, and few others I forget right now but that I will add later ;)
\n\nI'm happy on my side, my goal was to pass 0.8, I made it, barely, but I made it, with a very late submission that finished scoring after the deadline!
\n\nI may describe my approach in more detail in next few days, or later, depending on whether I chose to enter phase 2. My score is just 3 DBSCAN models with tracks merged when they overlap significantly. I don't use any track extension as my tracks are rather good in that respect. They rather contain too many outliers sometimes.
\n\nMy models focus on tracks originating from z axis. I spent the last few days trying to make DBSCAN work for secondary tracks. I got promising results but could not finish in time.
\n\nGiven organizers say they'll disclose prize winner code, I hope these winners will share their approach very soon in this forum... :)
\n\nedit: I shared my approach.
", "messages": [ { "id": 369894, "postDate": "2018-08-14T00:08:40.273Z", "content": "Congrats to all, esp the ones with amazing top scores, above or near 0.9. Let me aslo thank the organizers, CERN and Kaggle team for hosting such a challenge. I also want to thank those who shared so much here, including Yuval, Nicole Finnie, Heng Cher Keng, The Chemist (aka Grzegorz Sionkowski), Outrunner, Luis Andre, and few others I forget right now but that I will add later ;)
\n\nI'm happy on my side, my goal was to pass 0.8, I made it, barely, but I made it, with a very late submission that finished scoring after the deadline!
\n\nI may describe my approach in more detail in next few days, or later, depending on whether I chose to enter phase 2. My score is just 3 DBSCAN models with tracks merged when they overlap significantly. I don't use any track extension as my tracks are rather good in that respect. They rather contain too many outliers sometimes.
\n\nMy models focus on tracks originating from z axis. I spent the last few days trying to make DBSCAN work for secondary tracks. I got promising results but could not finish in time.
\n\nGiven organizers say they'll disclose prize winner code, I hope these winners will share their approach very soon in this forum... :)
\n\nedit: I shared my approach.
", "rawMarkdown": "Congrats to all, esp the ones with amazing top scores, above or near 0.9. Let me aslo thank the organizers, CERN and Kaggle team for hosting such a challenge. I also want to thank those who shared so much here, including Yuval, Nicole Finnie, Heng Cher Keng, The Chemist (aka Grzegorz Sionkowski), Outrunner, Luis Andre, and few others I forget right now but that I will add later ;)\n\nI'm happy on my side, my goal was to pass 0.8, I made it, barely, but I made it, with a very late submission that finished scoring after the deadline!\n\nI may describe my approach in more detail in next few days, or later, depending on whether I chose to enter phase 2. My score is just 3 DBSCAN models with tracks merged when they overlap significantly. I don't use any track extension as my tracks are rather good in that respect. They rather contain too many outliers sometimes. \n\nMy models focus on tracks originating from z axis. I spent the last few days trying to make DBSCAN work for secondary tracks. I got promising results but could not finish in time.\n\nGiven organizers say they'll disclose prize winner code, I hope these winners will share their approach very soon in this forum... :)\n\nedit: I [shared my approach][1].\n\n\n [1]: https://www.kaggle.com/c/trackml-particle-identification/discussion/63250#latest-369949", "votes": 13 }, { "id": 369899, "postDate": "2018-08-14T00:14:09.370Z", "content": "We (I and @Zidmie) are extremely glad to receive this gold medal. We have been watching the nearby teams so closely. Some shared feelings to the best silver of @Finnies. I'm sure, you are not so comfy to receive this silver like what you'd said. We were also in nervous mood in the final days. But don't forget to congratulate the last gold position, @Lonza, who made an impressive leap in the single final day. The same thought with @demelian, who lost some money prize in the last minute, and @Gorbunov, who created the most dramatic end to this competition.
", "rawMarkdown": "We (I and @Zidmie) are extremely glad to receive this gold medal. We have been watching the nearby teams so closely. Some shared feelings to the best silver of @Finnies. I'm sure, you are not so comfy to receive this silver like what you'd said. We were also in nervous mood in the final days. But don't forget to congratulate the last gold position, @Lonza, who made an impressive leap in the single final day. The same thought with @demelian, who lost some money prize in the last minute, and @Gorbunov, who created the most dramatic end to this competition.", "votes": 6 }, { "id": 369904, "postDate": "2018-08-14T00:21:58.997Z", "content": "CPMP, i congratulate you with fifth gold medal and Competition Grandmaster rank ))
", "rawMarkdown": "CPMP, i congratulate you with fifth gold medal and Competition Grandmaster rank ))", "votes": 3, "replies": [ { "id": 369905, "postDate": "2018-08-14T00:23:56.990Z", "content": "Thanks you, becoming competition GM was indeed a side bonus I was hoping for :)
", "rawMarkdown": "Thanks you, becoming competition GM was indeed a side bonus I was hoping for :)", "votes": 1 } ] }, { "id": 370371, "postDate": "2018-08-14T17:42:36.770Z", "content": "@CPMP, congratulations! Thank you for your much discussion. That's very usefull for me!
", "rawMarkdown": "@CPMP, congratulations! Thank you for your much discussion. That's very usefull for me!", "votes": 1 }, { "id": 370022, "postDate": "2018-08-14T05:23:45.757Z", "content": "Congratulations @CPMP and thanks for making the discussion threads lively and for all the bread crumbs.
", "rawMarkdown": "Congratulations @CPMP and thanks for making the discussion threads lively and for all the bread crumbs.", "votes": 1 }, { "id": 369960, "postDate": "2018-08-14T02:08:19.827Z", "content": "@CPMP, congratulation. I learned many things from your discussions and enjoyed the competition. thank you!
", "rawMarkdown": "@CPMP, congratulation. I learned many things from your discussions and enjoyed the competition. thank you! ", "votes": 1 }, { "id": 369925, "postDate": "2018-08-14T00:42:13.527Z", "content": "Congrats on gold and becoming Competitions Grandmaster and for everything you shared in the forums :)
", "rawMarkdown": "Congrats on gold and becoming Competitions Grandmaster and for everything you shared in the forums :)", "votes": 1 }, { "id": 369982, "postDate": "2018-08-14T03:29:13.253Z", "content": "Congratulations, and I am happy when I pass 0.9 too, but I think I should set it higher next time. :)
", "rawMarkdown": "Congratulations, and I am happy when I pass 0.9 too, but I think I should set it higher next time. :)", "votes": 2, "replies": [ { "id": 370161, "postDate": "2018-08-14T11:16:10.903Z", "content": "Yes, it is hard to stay motivated when you have such a lead in the competition. Issue is that it was possible to make progress without submitting. I know going in stealth mode was the way to go, but I don't like it.
", "rawMarkdown": "Yes, it is hard to stay motivated when you have such a lead in the competition. Issue is that it was possible to make progress without submitting. I know going in stealth mode was the way to go, but I don't like it." }, { "id": 370170, "postDate": "2018-08-14T11:42:21.163Z", "content": "But I always assume that someone can get 0.9 since there are so much paper/material shows CERN's method even lib and I wonder why \"real kagglers\" don't use it.
", "rawMarkdown": "But I always assume that someone can get 0.9 since there are so much paper/material shows CERN's method even lib and I wonder why \"real kagglers\" don't use it.", "votes": 1 } ] }, { "id": 369914, "postDate": "2018-08-14T00:31:18.033Z", "content": "Congrats - this was a fun comp - much more fun than I originally anticipated :) Thank you for your sharing throughout the competition.
", "rawMarkdown": "Congrats - this was a fun comp - much more fun than I originally anticipated :) Thank you for your sharing throughout the competition.", "votes": 2 }, { "id": 369919, "postDate": "2018-08-14T00:38:38.280Z", "rawMarkdown": "", "votes": 2, "isDeleted": true } ], "comments": [ { "id": 369899, "author_name": "Kha Vo", "author_url": "", "post_date": "2018-08-14T00:14:09.370000", "content": "We (I and @Zidmie) are extremely glad to receive this gold medal. We have been watching the nearby teams so closely. Some shared feelings to the best silver of @Finnies. I'm sure, you are not so comfy to receive this silver like what you'd said. We were also in nervous mood in the final days. But don't forget to congratulate the last gold position, @Lonza, who made an impressive leap in the single final day. The same thought with @demelian, who lost some money prize in the last minute, and @Gorbunov, who created the most dramatic end to this competition.
", "votes": 6, "replies": [] }, { "id": 369904, "author_name": "Alexander Kireev", "author_url": "", "post_date": "2018-08-14T00:21:58.997000", "content": "CPMP, i congratulate you with fifth gold medal and Competition Grandmaster rank ))
", "votes": 3, "replies": [ { "id": 369905, "author_name": "CPMP", "author_url": "", "post_date": "2018-08-14T00:23:56.990000", "content": "Thanks you, becoming competition GM was indeed a side bonus I was hoping for :)
", "votes": 1, "replies": [] } ] }, { "id": 370371, "author_name": "takus", "author_url": "", "post_date": "2018-08-14T17:42:36.770000", "content": "@CPMP, congratulations! Thank you for your much discussion. That's very usefull for me!
", "votes": 1, "replies": [] }, { "id": 370022, "author_name": "YaGana Sheriff-Hussaini", "author_url": "", "post_date": "2018-08-14T05:23:45.757000", "content": "Congratulations @CPMP and thanks for making the discussion threads lively and for all the bread crumbs.
", "votes": 1, "replies": [] }, { "id": 369960, "author_name": "Rei Matsuzaki", "author_url": "", "post_date": "2018-08-14T02:08:19.827000", "content": "@CPMP, congratulation. I learned many things from your discussions and enjoyed the competition. thank you!
", "votes": 1, "replies": [] }, { "id": 369925, "author_name": "Jack Vial", "author_url": "", "post_date": "2018-08-14T00:42:13.527000", "content": "Congrats on gold and becoming Competitions Grandmaster and for everything you shared in the forums :)
", "votes": 1, "replies": [] }, { "id": 369982, "author_name": "outrunner", "author_url": "", "post_date": "2018-08-14T03:29:13.253000", "content": "Congratulations, and I am happy when I pass 0.9 too, but I think I should set it higher next time. :)
", "votes": 2, "replies": [ { "id": 370161, "author_name": "CPMP", "author_url": "", "post_date": "2018-08-14T11:16:10.903000", "content": "Yes, it is hard to stay motivated when you have such a lead in the competition. Issue is that it was possible to make progress without submitting. I know going in stealth mode was the way to go, but I don't like it.
", "votes": 0, "replies": [] }, { "id": 370170, "author_name": "outrunner", "author_url": "", "post_date": "2018-08-14T11:42:21.163000", "content": "But I always assume that someone can get 0.9 since there are so much paper/material shows CERN's method even lib and I wonder why \"real kagglers\" don't use it.
", "votes": 1, "replies": [] } ] }, { "id": 369914, "author_name": "Μαριος Μιχαηλιδης KazAnova", "author_url": "", "post_date": "2018-08-14T00:31:18.033000", "content": "Congrats - this was a fun comp - much more fun than I originally anticipated :) Thank you for your sharing throughout the competition.
", "votes": 2, "replies": [] }, { "id": 369919, "author_name": "", "author_url": "", "post_date": "2018-08-14T00:38:38.280000", "content": "", "votes": 2, "replies": [] } ], "raw_markdown_by_id": { "369894": "Congrats to all, esp the ones with amazing top scores, above or near 0.9. Let me aslo thank the organizers, CERN and Kaggle team for hosting such a challenge. I also want to thank those who shared so much here, including Yuval, Nicole Finnie, Heng Cher Keng, The Chemist (aka Grzegorz Sionkowski), Outrunner, Luis Andre, and few others I forget right now but that I will add later ;)\n\nI'm happy on my side, my goal was to pass 0.8, I made it, barely, but I made it, with a very late submission that finished scoring after the deadline!\n\nI may describe my approach in more detail in next few days, or later, depending on whether I chose to enter phase 2. My score is just 3 DBSCAN models with tracks merged when they overlap significantly. I don't use any track extension as my tracks are rather good in that respect. They rather contain too many outliers sometimes. \n\nMy models focus on tracks originating from z axis. I spent the last few days trying to make DBSCAN work for secondary tracks. I got promising results but could not finish in time.\n\nGiven organizers say they'll disclose prize winner code, I hope these winners will share their approach very soon in this forum... :)\n\nedit: I [shared my approach][1].\n\n\n [1]: https://www.kaggle.com/c/trackml-particle-identification/discussion/63250#latest-369949", "369899": "We (I and @Zidmie) are extremely glad to receive this gold medal. We have been watching the nearby teams so closely. Some shared feelings to the best silver of @Finnies. I'm sure, you are not so comfy to receive this silver like what you'd said. We were also in nervous mood in the final days. But don't forget to congratulate the last gold position, @Lonza, who made an impressive leap in the single final day. The same thought with @demelian, who lost some money prize in the last minute, and @Gorbunov, who created the most dramatic end to this competition.", "369904": "CPMP, i congratulate you with fifth gold medal and Competition Grandmaster rank ))", "370371": "@CPMP, congratulations! Thank you for your much discussion. That's very usefull for me!", "370022": "Congratulations @CPMP and thanks for making the discussion threads lively and for all the bread crumbs.", "369960": "@CPMP, congratulation. I learned many things from your discussions and enjoyed the competition. thank you! ", "369925": "Congrats on gold and becoming Competitions Grandmaster and for everything you shared in the forums :)", "369982": "Congratulations, and I am happy when I pass 0.9 too, but I think I should set it higher next time. :)", "369914": "Congrats - this was a fun comp - much more fun than I originally anticipated :) Thank you for your sharing throughout the competition.", "369919": "" } }