What is a rerun / two stage competition:
\nTwo stage competitions a.k.a. reruns normally run over several reruns with of new data collected during the evaluation period, they usually also accumulate all the data starting after the deadline, with all the data running the final run. The first rerun often has a small part of data, here we can see an indicative score and that the solutions are running/working in the private part, but still the score will change during the race as more data are added.
\nUsed public datasets and notebooks:
\nCheck that all public data are is still there that the code is using. Also, don’t delete your own data during the long time of reruns and evaluation period (November 10).
Choosing submissions:
\nThe amount of final private test data should be at the same size as the public run but still maybe one selected submission shouldn’t be at the 8h+ stretch or close to the memory limits, not having all the submission gambling with time limit/memory failure, talking from own experience and thoughts. Of course, it’s a matter of personal choice and taste.
Try keeping the forum active, post some experience and thoughts during the evaluation period, after the deadline. That’s one of the values why we are at Kaggle, to learn, contribute and building the area. 👍
\nHappy Kaggle evaluation period! 🙂
", "messages": [ { "id": 1894105, "postDate": "2022-08-11T09:19:28.957Z", "content": "What is a rerun / two stage competition:
\nTwo stage competitions a.k.a. reruns normally run over several reruns with of new data collected during the evaluation period, they usually also accumulate all the data starting after the deadline, with all the data running the final run. The first rerun often has a small part of data, here we can see an indicative score and that the solutions are running/working in the private part, but still the score will change during the race as more data are added.
\nUsed public datasets and notebooks:
\nCheck that all public data are is still there that the code is using. Also, don’t delete your own data during the long time of reruns and evaluation period (November 10).
Choosing submissions:
\nThe amount of final private test data should be at the same size as the public run but still maybe one selected submission shouldn’t be at the 8h+ stretch or close to the memory limits, not having all the submission gambling with time limit/memory failure, talking from own experience and thoughts. Of course, it’s a matter of personal choice and taste.
Try keeping the forum active, post some experience and thoughts during the evaluation period, after the deadline. That’s one of the values why we are at Kaggle, to learn, contribute and building the area. 👍
\nHappy Kaggle evaluation period! 🙂
", "rawMarkdown": "What is a rerun / two stage competition:\n\nTwo stage competitions a.k.a. reruns normally run over several reruns with of new data collected during the evaluation period, they usually also accumulate all the data starting after the deadline, with all the data running the final run. The first rerun often has a small part of data, here we can see an indicative score and that the solutions are running/working in the private part, but still the score will change during the race as more data are added.\n\nUsed public datasets and notebooks: \nCheck that all public data are is still there that the code is using. Also, don’t delete your own data during the long time of reruns and evaluation period (November 10).\n\nChoosing submissions: \nThe amount of final private test data should be at the same size as the public run but still maybe one selected submission shouldn’t be at the 8h+ stretch or close to the memory limits, not having all the submission gambling with time limit/memory failure, talking from own experience and thoughts. Of course, it’s a matter of personal choice and taste.\n\nTry keeping the forum active, post some experience and thoughts during the evaluation period, after the deadline. That’s one of the values why we are at Kaggle, to learn, contribute and building the area. 👍\n\nHappy Kaggle evaluation period! 🙂", "votes": 12 }, { "id": 1895593, "postDate": "2022-08-12T08:57:27.793Z", "content": "@addisonhoward What to expect for the reruns schedule, every second week or once per month?
", "rawMarkdown": "@addisonhoward What to expect for the reruns schedule, every second week or once per month?", "replies": [ { "id": 1896062, "postDate": "2022-08-12T14:53:59.987Z", "content": "Hey @kirderf - the current plan is every two weeks or so
", "rawMarkdown": "Hey @kirderf - the current plan is every two weeks or so", "votes": 2 } ] }, { "id": 1894115, "postDate": "2022-08-11T09:27:15.187Z", "content": "Let's discuss top public solutions :) Very very interesting what people done
", "rawMarkdown": "Let's discuss top public solutions :) Very very interesting what people done", "replies": [ { "id": 1894122, "postDate": "2022-08-11T09:33:34.037Z", "content": "Yes let's, after the deadline, looking forward to it! :)
", "rawMarkdown": "Yes let's, after the deadline, looking forward to it! :)", "votes": 1 } ] }, { "id": 1894109, "postDate": "2022-08-11T09:22:51.310Z", "content": "The last line is absolutely important, we are all here to learn and develop throughout our Kaggle odyssey
", "rawMarkdown": "The last line is absolutely important, we are all here to learn and develop throughout our Kaggle odyssey", "replies": [ { "id": 1894118, "postDate": "2022-08-11T09:31:00.420Z", "content": "Indeed, Kaggle competitions is a community teamwork for a common problem and area, and an education class of high value, with a competition tail :)
", "rawMarkdown": "Indeed, Kaggle competitions is a community teamwork for a common problem and area, and an education class of high value, with a competition tail :)", "votes": 1 } ] }, { "id": 1895069, "postDate": "2022-08-12T00:04:33.030Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 1895593, "author_name": "Kirderf", "author_url": "", "post_date": "2022-08-12T08:57:27.793000", "content": "@addisonhoward What to expect for the reruns schedule, every second week or once per month?
", "votes": 0, "replies": [ { "id": 1896062, "author_name": "Addison Howard", "author_url": "", "post_date": "2022-08-12T14:53:59.987000", "content": "Hey @kirderf - the current plan is every two weeks or so
", "votes": 2, "replies": [] } ] }, { "id": 1894115, "author_name": "Vadim Irtlach", "author_url": "", "post_date": "2022-08-11T09:27:15.187000", "content": "Let's discuss top public solutions :) Very very interesting what people done
", "votes": 0, "replies": [ { "id": 1894122, "author_name": "Kirderf", "author_url": "", "post_date": "2022-08-11T09:33:34.037000", "content": "Yes let's, after the deadline, looking forward to it! :)
", "votes": 1, "replies": [] } ] }, { "id": 1894109, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2022-08-11T09:22:51.310000", "content": "The last line is absolutely important, we are all here to learn and develop throughout our Kaggle odyssey
", "votes": 0, "replies": [ { "id": 1894118, "author_name": "Kirderf", "author_url": "", "post_date": "2022-08-11T09:31:00.420000", "content": "Indeed, Kaggle competitions is a community teamwork for a common problem and area, and an education class of high value, with a competition tail :)
", "votes": 1, "replies": [] } ] }, { "id": 1895069, "author_name": "", "author_url": "", "post_date": "2022-08-12T00:04:33.030000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "1894105": "What is a rerun / two stage competition:\n\nTwo stage competitions a.k.a. reruns normally run over several reruns with of new data collected during the evaluation period, they usually also accumulate all the data starting after the deadline, with all the data running the final run. The first rerun often has a small part of data, here we can see an indicative score and that the solutions are running/working in the private part, but still the score will change during the race as more data are added.\n\nUsed public datasets and notebooks: \nCheck that all public data are is still there that the code is using. Also, don’t delete your own data during the long time of reruns and evaluation period (November 10).\n\nChoosing submissions: \nThe amount of final private test data should be at the same size as the public run but still maybe one selected submission shouldn’t be at the 8h+ stretch or close to the memory limits, not having all the submission gambling with time limit/memory failure, talking from own experience and thoughts. Of course, it’s a matter of personal choice and taste.\n\nTry keeping the forum active, post some experience and thoughts during the evaluation period, after the deadline. That’s one of the values why we are at Kaggle, to learn, contribute and building the area. 👍\n\nHappy Kaggle evaluation period! 🙂", "1895593": "@addisonhoward What to expect for the reruns schedule, every second week or once per month?", "1894115": "Let's discuss top public solutions :) Very very interesting what people done", "1894109": "The last line is absolutely important, we are all here to learn and develop throughout our Kaggle odyssey", "1895069": "" } }