With apologies to Douglas Adams!
\n\nI've had a great time working on this challenge for the last couple of months, and I've learned a lot.
\n\nReal life is intruding, and I won't have any time to continue trying to catch the leaders between now and the end of the competition.
\n\nI've got one more submission in the pipeline, but it will only get me to something between 0.66 and 0.67 - probably not enough to even advance one spot!
\n\nA HUGE thank you to the thought leaders on this board who have so generously contributed to my understanding and efforts:
\n\n@Heng\n@Grzegorz\n@CPMP\n@Nicole
\n\nI very much look forward seeing the approaches that are providing such amazing results - and figuring out why I was never able to find them ;^)
\n\nCheers - John
", "messages": [ { "id": 357669, "postDate": "2018-07-16T15:13:33.467Z", "content": "With apologies to Douglas Adams!
\n\nI've had a great time working on this challenge for the last couple of months, and I've learned a lot.
\n\nReal life is intruding, and I won't have any time to continue trying to catch the leaders between now and the end of the competition.
\n\nI've got one more submission in the pipeline, but it will only get me to something between 0.66 and 0.67 - probably not enough to even advance one spot!
\n\nA HUGE thank you to the thought leaders on this board who have so generously contributed to my understanding and efforts:
\n\n@Heng\n@Grzegorz\n@CPMP\n@Nicole
\n\nI very much look forward seeing the approaches that are providing such amazing results - and figuring out why I was never able to find them ;^)
\n\nCheers - John
", "rawMarkdown": "With apologies to Douglas Adams!\n\nI've had a great time working on this challenge for the last couple of months, and I've learned a lot.\n\nReal life is intruding, and I won't have any time to continue trying to catch the leaders between now and the end of the competition.\n\nI've got one more submission in the pipeline, but it will only get me to something between 0.66 and 0.67 - probably not enough to even advance one spot!\n\nA HUGE thank you to the thought leaders on this board who have so generously contributed to my understanding and efforts:\n\n@Heng\n@Grzegorz\n@CPMP\n@Nicole\n\nI very much look forward seeing the approaches that are providing such amazing results - and figuring out why I was never able to find them ;^)\n\nCheers - John", "votes": 8 }, { "id": 357766, "postDate": "2018-07-16T19:57:43.887Z", "content": "Thanks, and good luck for whatever takes you away from this.
", "rawMarkdown": "Thanks, and good luck for whatever takes you away from this.", "votes": 1, "replies": [ { "id": 357785, "postDate": "2018-07-16T21:00:57.500Z", "content": "Oh, it's nothing serious. A family vacation with no HW attached 😁
", "rawMarkdown": "Oh, it's nothing serious. A family vacation with no HW attached 😁", "votes": 2 }, { "id": 357890, "postDate": "2018-07-17T05:00:19.320Z", "content": "Even better. Enjoy!
", "rawMarkdown": "Even better. Enjoy!" }, { "id": 359323, "postDate": "2018-07-19T21:38:59.943Z", "content": "@John, have a nice vacation (What? Are you gonna go away for a month? Since when did people get so many vacation days in the US :p I thought it's only for us lazy Europeans). I took a long break from this competition, since our score is way off from top scorers, we're just exploring new approaches for fun. :)
", "rawMarkdown": "@John, have a nice vacation (What? Are you gonna go away for a month? Since when did people get so many vacation days in the US :p I thought it's only for us lazy Europeans). I took a long break from this competition, since our score is way off from top scorers, we're just exploring new approaches for fun. :)" }, { "id": 359595, "postDate": "2018-07-20T12:20:32.740Z", "content": "@Nicole - I haven't gone full European... Really just gone for ~3 weeks! Not much time to do anything new when I return - but maybe I can do something with the Chemist's last insight about unused hits and tracks with even numbers... Thanks
", "rawMarkdown": "@Nicole - I haven't gone full European... Really just gone for ~3 weeks! Not much time to do anything new when I return - but maybe I can do something with the Chemist's last insight about unused hits and tracks with even numbers... Thanks", "votes": 1 } ] }, { "id": 357685, "postDate": "2018-07-16T15:45:17.617Z", "content": "congrats mate keep on going
", "rawMarkdown": "congrats mate keep on going", "votes": 1 }, { "id": 358332, "postDate": "2018-07-18T00:37:35.147Z", "content": "I've only just joined kaggle so am as green as grass, and won't have time to do this before August 6th, but do you know if in kaggle they make the truth files for the test available after the end of the competition, so I can see what my score would have been when I get a chance to run a model?
", "rawMarkdown": "I've only just joined kaggle so am as green as grass, and won't have time to do this before August 6th, but do you know if in kaggle they make the truth files for the test available after the end of the competition, so I can see what my score would have been when I get a chance to run a model?", "replies": [ { "id": 358346, "postDate": "2018-07-18T01:25:33.737Z", "content": "I think sometimes they do but it is up to the sponsors. For example in the 2018 Data Science Bowl contest, the sponsors asked the community if they would like that to happen. See the link below:-\nhttps://www.kaggle.com/c/data-science-bowl-2018/discussion/54764
", "rawMarkdown": "I think sometimes they do but it is up to the sponsors. For example in the 2018 Data Science Bowl contest, the sponsors asked the community if they would like that to happen. See the link below:-\nhttps://www.kaggle.com/c/data-science-bowl-2018/discussion/54764" }, { "id": 358375, "postDate": "2018-07-18T03:33:37.627Z", "content": "for this competition you don't need the ground truth for test data as there is a pretty good correlation between your score on train data and the leaderboard. There is way enough event data in the trian set to properly evaluate your models.
\n\nThis may not be true for other competitions.
", "rawMarkdown": "for this competition you don't need the ground truth for test data as there is a pretty good correlation between your score on train data and the leaderboard. There is way enough event data in the trian set to properly evaluate your models. \n\nThis may not be true for other competitions." } ] }, { "id": 358489, "postDate": "2018-07-18T08:27:25.397Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 357766, "author_name": "CPMP", "author_url": "", "post_date": "2018-07-16T19:57:43.887000", "content": "Thanks, and good luck for whatever takes you away from this.
", "votes": 1, "replies": [ { "id": 357785, "author_name": "John Sweeney", "author_url": "", "post_date": "2018-07-16T21:00:57.500000", "content": "Oh, it's nothing serious. A family vacation with no HW attached 😁
", "votes": 2, "replies": [] }, { "id": 357890, "author_name": "CPMP", "author_url": "", "post_date": "2018-07-17T05:00:19.320000", "content": "Even better. Enjoy!
", "votes": 0, "replies": [] }, { "id": 359323, "author_name": "Nicole Finnie", "author_url": "", "post_date": "2018-07-19T21:38:59.943000", "content": "@John, have a nice vacation (What? Are you gonna go away for a month? Since when did people get so many vacation days in the US :p I thought it's only for us lazy Europeans). I took a long break from this competition, since our score is way off from top scorers, we're just exploring new approaches for fun. :)
", "votes": 0, "replies": [] }, { "id": 359595, "author_name": "John Sweeney", "author_url": "", "post_date": "2018-07-20T12:20:32.740000", "content": "@Nicole - I haven't gone full European... Really just gone for ~3 weeks! Not much time to do anything new when I return - but maybe I can do something with the Chemist's last insight about unused hits and tracks with even numbers... Thanks
", "votes": 1, "replies": [] } ] }, { "id": 357685, "author_name": "dineshbarri", "author_url": "", "post_date": "2018-07-16T15:45:17.617000", "content": "congrats mate keep on going
", "votes": 1, "replies": [] }, { "id": 358332, "author_name": "Raymond Gabriel", "author_url": "", "post_date": "2018-07-18T00:37:35.147000", "content": "I've only just joined kaggle so am as green as grass, and won't have time to do this before August 6th, but do you know if in kaggle they make the truth files for the test available after the end of the competition, so I can see what my score would have been when I get a chance to run a model?
", "votes": 0, "replies": [ { "id": 358346, "author_name": "YaGana Sheriff-Hussaini", "author_url": "", "post_date": "2018-07-18T01:25:33.737000", "content": "I think sometimes they do but it is up to the sponsors. For example in the 2018 Data Science Bowl contest, the sponsors asked the community if they would like that to happen. See the link below:-\nhttps://www.kaggle.com/c/data-science-bowl-2018/discussion/54764
", "votes": 0, "replies": [] }, { "id": 358375, "author_name": "CPMP", "author_url": "", "post_date": "2018-07-18T03:33:37.627000", "content": "for this competition you don't need the ground truth for test data as there is a pretty good correlation between your score on train data and the leaderboard. There is way enough event data in the trian set to properly evaluate your models.
\n\nThis may not be true for other competitions.
", "votes": 0, "replies": [] } ] }, { "id": 358489, "author_name": "", "author_url": "", "post_date": "2018-07-18T08:27:25.397000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "357669": "With apologies to Douglas Adams!\n\nI've had a great time working on this challenge for the last couple of months, and I've learned a lot.\n\nReal life is intruding, and I won't have any time to continue trying to catch the leaders between now and the end of the competition.\n\nI've got one more submission in the pipeline, but it will only get me to something between 0.66 and 0.67 - probably not enough to even advance one spot!\n\nA HUGE thank you to the thought leaders on this board who have so generously contributed to my understanding and efforts:\n\n@Heng\n@Grzegorz\n@CPMP\n@Nicole\n\nI very much look forward seeing the approaches that are providing such amazing results - and figuring out why I was never able to find them ;^)\n\nCheers - John", "357766": "Thanks, and good luck for whatever takes you away from this.", "357685": "congrats mate keep on going", "358332": "I've only just joined kaggle so am as green as grass, and won't have time to do this before August 6th, but do you know if in kaggle they make the truth files for the test available after the end of the competition, so I can see what my score would have been when I get a chance to run a model?", "358489": "" } }