There are still 3 days left until this competition. Do you think this competition will shake?
", "messages": [ { "id": 3056435, "postDate": "2024-11-27T00:54:19.787Z", "content": "There are still 3 days left until this competition. Do you think this competition will shake?
", "rawMarkdown": "There are still 3 days left until this competition. Do you think this competition will shake?", "votes": 7 }, { "id": 3058138, "postDate": "2024-11-29T05:24:31.763Z", "content": "Given the relationship between CV and LB it seems very likely. The main question is probably not IF there will be a shakeup, but rather how bad it will be.
", "rawMarkdown": "Given the relationship between CV and LB it seems very likely. The main question is probably not IF there will be a shakeup, but rather how bad it will be. ", "votes": 1, "replies": [ { "id": 3058162, "postDate": "2024-11-29T06:22:50.410Z", "content": "True, my CV is a bit correlated with the LB, but fluctuates a lot - this is a good precursor for a nice shakeup @thomasmeiner
", "rawMarkdown": "True, my CV is a bit correlated with the LB, but fluctuates a lot - this is a good precursor for a nice shakeup @thomasmeiner ", "votes": 1 } ] }, { "id": 3056566, "postDate": "2024-11-27T05:38:05.313Z", "content": "I think yes, from my limited CV-LB relation insofar @yunsuxiaozi
", "rawMarkdown": "I think yes, from my limited CV-LB relation insofar @yunsuxiaozi ", "votes": 1 }, { "id": 3058502, "postDate": "2024-11-29T14:07:22.117Z", "content": "This it's my first official competition on kaggle, so I could learn all the day new things.
\nThanks for asking, I did not notice that yet and I think that yes 😱😱😱! There will be very likely a new sort order of the winners.
\nDo I see this right, that all submissions done by me or the group are counting, right? So if I submitted every day 3 times, testing different approaches, the possibility will also be higher to get higher in the final leaderboard?! @thomasmeiner @ravi20076 @tilii7
", "rawMarkdown": "This it's my first official competition on kaggle, so I could learn all the day new things. \n\nThanks for asking, I did not notice that yet and I think that yes 😱😱😱! There will be very likely a new sort order of the winners.\n\nDo I see this right, that all submissions done by me or the group are counting, right? So if I submitted every day 3 times, testing different approaches, the possibility will also be higher to get higher in the final leaderboard?! @thomasmeiner @ravi20076 @tilii7 ", "votes": 2, "replies": [ { "id": 3058591, "postDate": "2024-11-29T16:25:12.700Z", "content": "All your submissions will count but you need to choose a submission for the private lb. Else kaggle will select the best public score submission as the final one for the private lb.
\nThe number of experiments you do definitely helps you with a better chance at a good model and eventually a better score, but luck and choice of submission also matter.
\nAll the best @ralfjunghanns
", "rawMarkdown": "All your submissions will count but you need to choose a submission for the private lb. Else kaggle will select the best public score submission as the final one for the private lb.\n\nThe number of experiments you do definitely helps you with a better chance at a good model and eventually a better score, but luck and choice of submission also matter. \n\nAll the best @ralfjunghanns ", "votes": 1 }, { "id": 3058723, "postDate": "2024-11-29T20:53:46.590Z", "content": "On your submissions page there is a vertical blue stripe on the right side that indicates which submission will be automatically selected for private LB scoring. If you don't select a different submission by clicking on its right-side box, that submission will be the one that determines your private LB placement.
\nLike Ravi told you, all the submissions will get private LB scores, but only the selected submission will count for the final LB placement. The whole trick is to select the submission you think will generalize the best, which may or may not be the submission with the best public LB score.
", "rawMarkdown": "On your submissions page there is a vertical blue stripe on the right side that indicates which submission will be automatically selected for private LB scoring. If you don't select a different submission by clicking on its right-side box, that submission will be the one that determines your private LB placement.\n\nLike Ravi told you, all the submissions will get private LB scores, but only the selected submission will count for the final LB placement. The whole trick is to select the submission you think will generalize the best, which may or may not be the submission with the best public LB score." }, { "id": 3070701, "postDate": "2024-12-13T01:02:02.743Z", "content": "my first participation never happened and i want to ask a few questions to the newcomer who took 10th place.
", "rawMarkdown": "my first participation never happened and i want to ask a few questions to the newcomer who took 10th place.", "replies": [ { "id": 3070873, "postDate": "2024-12-13T06:02:05.547Z", "content": "Feel free to ask. Send me a DM.
", "rawMarkdown": "Feel free to ask. Send me a DM." } ] }, { "id": 3070703, "postDate": "2024-12-13T01:04:58.747Z", "content": "why only two forecasting models are used? can i use my own approach? the main thing is to provide the forecasting results?
", "rawMarkdown": "why only two forecasting models are used? can i use my own approach? the main thing is to provide the forecasting results?", "replies": [ { "id": 3070871, "postDate": "2024-12-13T06:00:53.790Z", "rawMarkdown": "", "isDeleted": true }, { "id": 3070874, "postDate": "2024-12-13T06:03:35.397Z", "content": "I think you can use the approach you want if it's able to predict \"bg+1:00\" based on the lag features.
", "rawMarkdown": "I think you can use the approach you want if it's able to predict \"bg+1:00\" based on the lag features." }, { "id": 3070897, "postDate": "2024-12-13T06:35:20.660Z", "rawMarkdown": "", "isDeleted": true } ] } ] }, { "id": 3056458, "postDate": "2024-11-27T01:43:39.990Z", "content": "Given a spread between CV and public LB, plus a single submission, it would be surprising if it doesn't.
", "rawMarkdown": "Given a spread between CV and public LB, plus a single submission, it would be surprising if it doesn't.", "votes": 2 }, { "id": 3058773, "postDate": "2024-11-29T23:43:09.350Z", "content": "There will likely be a shakeup. Medical data and sensor data is noisy, making it extra prone to overfitting. It would be surprising if there wasn't a shakeup.
", "rawMarkdown": "There will likely be a shakeup. Medical data and sensor data is noisy, making it extra prone to overfitting. It would be surprising if there wasn't a shakeup." } ], "comments": [ { "id": 3058138, "author_name": "Thomas Meißner", "author_url": "", "post_date": "2024-11-29T05:24:31.763000", "content": "Given the relationship between CV and LB it seems very likely. The main question is probably not IF there will be a shakeup, but rather how bad it will be.
", "votes": 1, "replies": [ { "id": 3058162, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-11-29T06:22:50.410000", "content": "True, my CV is a bit correlated with the LB, but fluctuates a lot - this is a good precursor for a nice shakeup @thomasmeiner
", "votes": 1, "replies": [] } ] }, { "id": 3056566, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-11-27T05:38:05.313000", "content": "I think yes, from my limited CV-LB relation insofar @yunsuxiaozi
", "votes": 1, "replies": [] }, { "id": 3058502, "author_name": "Ralf Junghanns", "author_url": "", "post_date": "2024-11-29T14:07:22.117000", "content": "This it's my first official competition on kaggle, so I could learn all the day new things.
\nThanks for asking, I did not notice that yet and I think that yes 😱😱😱! There will be very likely a new sort order of the winners.
\nDo I see this right, that all submissions done by me or the group are counting, right? So if I submitted every day 3 times, testing different approaches, the possibility will also be higher to get higher in the final leaderboard?! @thomasmeiner @ravi20076 @tilii7
", "votes": 2, "replies": [ { "id": 3058591, "author_name": "Ravi Ramakrishnan", "author_url": "", "post_date": "2024-11-29T16:25:12.700000", "content": "All your submissions will count but you need to choose a submission for the private lb. Else kaggle will select the best public score submission as the final one for the private lb.
\nThe number of experiments you do definitely helps you with a better chance at a good model and eventually a better score, but luck and choice of submission also matter.
\nAll the best @ralfjunghanns
", "votes": 1, "replies": [] }, { "id": 3058723, "author_name": "Tilii", "author_url": "", "post_date": "2024-11-29T20:53:46.590000", "content": "On your submissions page there is a vertical blue stripe on the right side that indicates which submission will be automatically selected for private LB scoring. If you don't select a different submission by clicking on its right-side box, that submission will be the one that determines your private LB placement.
\nLike Ravi told you, all the submissions will get private LB scores, but only the selected submission will count for the final LB placement. The whole trick is to select the submission you think will generalize the best, which may or may not be the submission with the best public LB score.
", "votes": 0, "replies": [] }, { "id": 3070701, "author_name": "Сергей Шишкин", "author_url": "", "post_date": "2024-12-13T01:02:02.743000", "content": "my first participation never happened and i want to ask a few questions to the newcomer who took 10th place.
", "votes": 0, "replies": [ { "id": 3070873, "author_name": "Ralf Junghanns", "author_url": "", "post_date": "2024-12-13T06:02:05.547000", "content": "Feel free to ask. Send me a DM.
", "votes": 0, "replies": [] } ] }, { "id": 3070703, "author_name": "Сергей Шишкин", "author_url": "", "post_date": "2024-12-13T01:04:58.747000", "content": "why only two forecasting models are used? can i use my own approach? the main thing is to provide the forecasting results?
", "votes": 0, "replies": [ { "id": 3070871, "author_name": "", "author_url": "", "post_date": "2024-12-13T06:00:53.790000", "content": "", "votes": 0, "replies": [] }, { "id": 3070874, "author_name": "Ralf Junghanns", "author_url": "", "post_date": "2024-12-13T06:03:35.397000", "content": "I think you can use the approach you want if it's able to predict \"bg+1:00\" based on the lag features.
", "votes": 0, "replies": [] }, { "id": 3070897, "author_name": "", "author_url": "", "post_date": "2024-12-13T06:35:20.660000", "content": "", "votes": 0, "replies": [] } ] } ] }, { "id": 3056458, "author_name": "Tilii", "author_url": "", "post_date": "2024-11-27T01:43:39.990000", "content": "Given a spread between CV and public LB, plus a single submission, it would be surprising if it doesn't.
", "votes": 2, "replies": [] }, { "id": 3058773, "author_name": "Josh Haber", "author_url": "", "post_date": "2024-11-29T23:43:09.350000", "content": "There will likely be a shakeup. Medical data and sensor data is noisy, making it extra prone to overfitting. It would be surprising if there wasn't a shakeup.
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "3056435": "There are still 3 days left until this competition. Do you think this competition will shake?", "3058138": "Given the relationship between CV and LB it seems very likely. The main question is probably not IF there will be a shakeup, but rather how bad it will be. ", "3056566": "I think yes, from my limited CV-LB relation insofar @yunsuxiaozi ", "3058502": "This it's my first official competition on kaggle, so I could learn all the day new things. \n\nThanks for asking, I did not notice that yet and I think that yes 😱😱😱! There will be very likely a new sort order of the winners.\n\nDo I see this right, that all submissions done by me or the group are counting, right? So if I submitted every day 3 times, testing different approaches, the possibility will also be higher to get higher in the final leaderboard?! @thomasmeiner @ravi20076 @tilii7 ", "3056458": "Given a spread between CV and public LB, plus a single submission, it would be surprising if it doesn't.", "3058773": "There will likely be a shakeup. Medical data and sensor data is noisy, making it extra prone to overfitting. It would be surprising if there wasn't a shakeup." } }