The goal of this competition is to use seismic signals to predict the timing of laboratory earthquakes. The data comes from a well-known experimental set-up used to study earthquake physics. Can the Kaggle community can help push the boundaries of this scientific research?
\n\nIf you want to get started fast, this Basic Feature Benchmark Kernel might be helpful.
\n\nPlease feel free to ask your questions in this thread.
\n\nGood luck!
", "messages": [ { "id": 2142772, "postDate": "2023-02-13T19:03:26.573Z", "content": "Hi. I'm confused about feature engineering. why do we take the last value of the target for every segment (150000).
", "rawMarkdown": "Hi. I'm confused about feature engineering. why do we take the last value of the target for every segment (150000).", "votes": 1 }, { "id": 453856, "postDate": "2019-01-10T22:40:54.603Z", "content": "The goal of this competition is to use seismic signals to predict the timing of laboratory earthquakes. The data comes from a well-known experimental set-up used to study earthquake physics. Can the Kaggle community can help push the boundaries of this scientific research?
\n\nIf you want to get started fast, this Basic Feature Benchmark Kernel might be helpful.
\n\nPlease feel free to ask your questions in this thread.
\n\nGood luck!
", "rawMarkdown": "The goal of this competition is to use seismic signals to predict the timing of laboratory earthquakes. The data comes from a well-known experimental set-up used to study earthquake physics. Can the Kaggle community can help push the boundaries of this scientific research?\n\nIf you want to get started fast, this [Basic Feature Benchmark Kernel](https://www.kaggle.com/inversion/basic-feature-benchmark) might be helpful.\n\nPlease feel free to ask your questions in this thread.\n\nGood luck!", "votes": 43 }, { "id": 454546, "postDate": "2019-01-11T20:16:13.547Z", "content": "Hi. This looks interesting.
\n\nI have a methodological, well, not \"quibble\", but suggestion for the future, at any rate: forecasts tend to be more valuable and actionable if they are probabilistic, rather than point-valued. That is, a full pdf of the time-to-failure (in your application) would be more useful for making decisions under uncertainty about the future than would a single summary statistic from the pdf, such as the mean or the mode. In weather forecasting (and in other applications, such as finance, epidemiology, etc.) this style of forecasting is common -- the forecast is a pdf conditioned on current information. Not only does this enable more powerful forecast assessment skill scores, but it also makes possible more powerful risk management methodologies.
", "rawMarkdown": "Hi. This looks interesting.\n\nI have a methodological, well, not \"quibble\", but suggestion for the future, at any rate: forecasts tend to be more valuable and actionable if they are probabilistic, rather than point-valued. That is, a full pdf of the time-to-failure (in your application) would be more useful for making decisions under uncertainty about the future than would a single summary statistic from the pdf, such as the mean or the mode. In weather forecasting (and in other applications, such as finance, epidemiology, etc.) this style of forecasting is common -- the forecast is a pdf conditioned on current information. Not only does this enable more powerful forecast assessment skill scores, but it also makes possible more powerful risk management methodologies.", "votes": 14 }, { "id": 519203, "postDate": "2019-04-18T14:37:26.280Z", "content": "Hi I didn't understand what is meant by predicting the remaining time. Is it the remaining time between current earthquake and occurrence of next earthquake ? Also I'm new to Kaggle. Can I use python tensorflow for this implementation ?
", "rawMarkdown": "Hi I didn't understand what is meant by predicting the remaining time. Is it the remaining time between current earthquake and occurrence of next earthquake ? Also I'm new to Kaggle. Can I use python tensorflow for this implementation ?", "votes": 7, "replies": [ { "id": 533955, "postDate": "2019-05-20T10:22:43.133Z", "content": "Yes, it the remaining time between the current earthquake and the occurrence of the next earthquake. And, you can either use R or Python.
", "rawMarkdown": "Yes, it the remaining time between the current earthquake and the occurrence of the next earthquake. And, you can either use R or Python." } ] }, { "id": 454381, "postDate": "2019-01-11T14:58:14.020Z", "content": "Is MATLAB acceptable for the competition? If not, would I need to use a free trial license to MATLAB issued to me for the express use of competing on this project?
", "rawMarkdown": "Is MATLAB acceptable for the competition? If not, would I need to use a free trial license to MATLAB issued to me for the express use of competing on this project?", "votes": 1, "replies": [ { "id": 454481, "postDate": "2019-01-11T17:48:12.350Z", "content": "Hi Chris,
\n\nPer the rules of the competition, all solutions must be open-sourced. As Matlab would require a paid license for the host to use, it would not be considered an open-sourced solutions.
\n\nThanks!
", "rawMarkdown": "Hi Chris,\n\nPer the rules of the competition, all solutions must be open-sourced. As Matlab would require a paid license for the host to use, it would not be considered an open-sourced solutions.\n\nThanks!", "votes": 4 }, { "id": 455780, "postDate": "2019-01-14T15:25:59.373Z", "content": "As a curiosity, would Octave code be acceptable then?
", "rawMarkdown": "As a curiosity, would Octave code be acceptable then? ", "votes": 4 } ] }, { "id": 456405, "postDate": "2019-01-15T18:47:21.643Z", "content": "I believe that the data must be 'chunked' by the earthquake events. Is it true that anything that happened at cycle k, (cycle is time between observations, where time-to-event=0), should not matter to then next cycle k+1? I wonder if test data sampled strictly from the 'before event' window.
", "rawMarkdown": "I believe that the data must be 'chunked' by the earthquake events. Is it true that anything that happened at cycle k, (cycle is time between observations, where time-to-event=0), should not matter to then next cycle k+1? I wonder if test data sampled strictly from the 'before event' window. ", "votes": 2 }, { "id": 454749, "postDate": "2019-01-12T06:10:40.447Z", "content": "Hello Addison.
\n\nMathworks (makers of MATLAB) has hosted Kaggle competitions before and as part of their sponsorship, they have provided free licenses to MATLAB and any necessary toolboxes for the duration of the competition. Their algorithms are not hidden - they are available for the user to peruse. Would this be an option for this competition? I spoke with a MATLAB representative this morning and he was interested in working something out so that myself and any other users could freely access MATLAB for the purpose of the competition. BTW, I have seen Q&As with former Kaggle personnel where MATLAB was allowed. Are you absolutely certain using MATLAB is banned from being an acceptable entry for a winning solution?
\n\nHere is an example where MATLAB was used: https://www.kaggle.com/c/melbourne-university-seizure-prediction
\n\nBest regards\nChris
", "rawMarkdown": "Hello Addison.\n\nMathworks (makers of MATLAB) has hosted Kaggle competitions before and as part of their sponsorship, they have provided free licenses to MATLAB and any necessary toolboxes for the duration of the competition. Their algorithms are not hidden - they are available for the user to peruse. Would this be an option for this competition? I spoke with a MATLAB representative this morning and he was interested in working something out so that myself and any other users could freely access MATLAB for the purpose of the competition. BTW, I have seen Q&As with former Kaggle personnel where MATLAB was allowed. Are you absolutely certain using MATLAB is banned from being an acceptable entry for a winning solution?\n\nHere is an example where MATLAB was used: https://www.kaggle.com/c/melbourne-university-seizure-prediction\n\nBest regards\nChris", "votes": 2, "replies": [ { "id": 454752, "postDate": "2019-01-12T06:16:22.870Z", "content": "More info about where MATLAB was an acceptable for a Kaggle competition: https://www.kaggle.com/c/noaa-right-whale-recognition/discussion/16161#91048
", "rawMarkdown": "More info about where MATLAB was an acceptable for a Kaggle competition: https://www.kaggle.com/c/noaa-right-whale-recognition/discussion/16161#91048\n" }, { "id": 455132, "postDate": "2019-01-13T03:20:37.747Z", "content": "Hi Chris, Thank you for your information. I am interested to use Matlab too. Please let me know Mathworks decision. I have used Matlab since early 90s and bought commercial license myself 10 years ago - using matlab would make things easier if the NN toolbox is included. The kaggle users can really do them a favor to test the toolbox. I am seriously considering to move to Python because of progress of Jupyter Notebook. Again thanks.
", "rawMarkdown": "Hi Chris, Thank you for your information. I am interested to use Matlab too. Please let me know Mathworks decision. I have used Matlab since early 90s and bought commercial license myself 10 years ago - using matlab would make things easier if the NN toolbox is included. The kaggle users can really do them a favor to test the toolbox. I am seriously considering to move to Python because of progress of Jupyter Notebook. Again thanks." } ] }, { "id": 1348762, "postDate": "2021-06-14T08:25:59.623Z", "content": "Hello! Could you please tell me why in the trainig data we have time_to_failure for each acoustic signal, but in the result we have one value of time_to_failure for 150 000 values of acoustic data? How can I compare my result with what is given?
", "rawMarkdown": "Hello! Could you please tell me why in the trainig data we have time_to_failure for each acoustic signal, but in the result we have one value of time_to_failure for 150 000 values of acoustic data? How can I compare my result with what is given?" }, { "id": 710205, "postDate": "2020-01-04T12:59:08.547Z", "content": "That was a great competition!
", "rawMarkdown": "That was a great competition! " }, { "id": 558277, "postDate": "2019-06-22T04:38:27.457Z", "content": "Hi, Today i have submitted my result file of LANLEarthquake Prediction, But I am not getting my name in Leaderboard,\nI do not understand why? Can anyone tell me?
", "rawMarkdown": "Hi, Today i have submitted my result file of LANLEarthquake Prediction, But I am not getting my name in Leaderboard,\nI do not understand why? Can anyone tell me?" }, { "id": 531627, "postDate": "2019-05-15T08:22:53.470Z", "content": "nice.
", "rawMarkdown": "nice." }, { "id": 531623, "postDate": "2019-05-15T08:08:16.120Z", "content": "Help! I could not load the LANL data set in my kernel.\nCould the Kaggle Team check the dataset?\nThanks!
", "rawMarkdown": "Help! I could not load the LANL data set in my kernel.\nCould the Kaggle Team check the dataset?\nThanks!" }, { "id": 519186, "postDate": "2019-04-18T14:14:32.577Z", "content": "Anyone else facing 404 - Not Found problem while downloading datasets via kaggle api command ?\nEDIT: Just checked other competitions' kaggle api command and they are giving same error as well.
just in case you hadn't come across the solution and it's the same thing I was facing on colab, changing over to a CPU kernel to download the data did the trick
", "rawMarkdown": "just in case you hadn't come across the solution and it's the same thing I was facing on colab, changing over to a CPU kernel to download the data did the trick" } ] }, { "id": 492377, "postDate": "2019-03-17T07:00:35.873Z", "content": "Interesting problem statement, especially for those who enjoy physics. I have used Svm SVR for this, and got 1.69 score . How can I improve my model, any suggestions?
", "rawMarkdown": "Interesting problem statement, especially for those who enjoy physics. I have used Svm SVR for this, and got 1.69 score . How can I improve my model, any suggestions?" }, { "id": 480609, "postDate": "2019-02-28T12:15:22.663Z", "content": "Hi!\nCould you please clarify, what does that mean: \"External Data Disclosure deadline. All external data used in the competition must be disclosed in the forums by this date.\" - What is \"external data\" in this particular case?\nAlso, am I right assuming that there'll be no private leaderboard and no additional data for final scoring?
\n\np.s. sorry for dumb questions, this competition is baptism of fire on kaggle :)\nThanks!
", "rawMarkdown": "Hi!\nCould you please clarify, what does that mean: \"External Data Disclosure deadline. All external data used in the competition must be disclosed in the forums by this date.\" - What is \"external data\" in this particular case?\nAlso, am I right assuming that there'll be no private leaderboard and no additional data for final scoring?\n\np.s. sorry for dumb questions, this competition is baptism of fire on kaggle :)\nThanks!", "replies": [ { "id": 480808, "postDate": "2019-02-28T17:36:42.197Z", "content": "Hi Stanislav,
\n\nIf you use any data or pre-trained models not provided on the competition Data page, you must disclose what they are before the competition data deadline.
\n\nThere is a private leaderboard, but it is a subset of the full test set.
", "rawMarkdown": "Hi Stanislav,\n\nIf you use any data or pre-trained models not provided on the competition Data page, you must disclose what they are before the competition data deadline.\n\nThere is a private leaderboard, but it is a subset of the full test set.", "votes": 2 }, { "id": 480817, "postDate": "2019-02-28T17:58:28.543Z", "content": "Thanks for your reply!
", "rawMarkdown": "Thanks for your reply!" }, { "id": 536918, "postDate": "2019-05-25T16:29:13.870Z", "content": "@addisonhoward @inversion Can we use external data or pre-trained model that has been disclosed/mentioned by other people during the competition, or is it an individual disclosure letting the kaggle team and others you are using external data/pre-trained models. Hoping for last minute clarification. Thanks!
", "rawMarkdown": "@addisonhoward @inversion Can we use external data or pre-trained model that has been disclosed/mentioned by other people during the competition, or is it an individual disclosure letting the kaggle team and others you are using external data/pre-trained models. Hoping for last minute clarification. Thanks!" }, { "id": 537099, "postDate": "2019-05-26T07:13:51.843Z", "content": "@teeyee314 This is covered in the competition rules, external data section. You have to disclose what you use in the competition forum at least one week before competition end:
\n\n\n", "rawMarkdown": "@teeyee314 This is covered in the competition rules, external data section. You have to disclose what you use in the competition forum at least one week before competition end:\n\n> EXTERNAL DATA\n\n> The following provision supersedes General Rules Section 7.C. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.\"", "votes": 1 }, { "id": 537111, "postDate": "2019-05-26T08:01:02.550Z", "content": "EXTERNAL DATA
\n \nThe following provision supersedes General Rules Section 7.C. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.\"
\n
@cpmpml I read that thoroughly. It is not clear if I need publicly say that I am using the external data shared by other people in the discussion. To me rule is not clear about that. But I guess to be safe, I should just make a disclosure. No harm in doing that.
\n\n> provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.
\n\nthere is no \"disclosure rule\", merely a \"fairness clause\". This is how I'm interpreting it
\n\nPlease use this thread to list the pre-trained models you are planning to use, or other external data. Once a model or data source has been posted once, it does not need to be posted again. You must post here one week prior to competition close. -Inversion
since the external data I'm planning on using is already posted per the \"fairness clause\", I see no need to individual state that I am using it.
", "rawMarkdown": "@cpmpml I read that thoroughly. It is not clear if I need publicly say that I am using the external data shared by other people in the discussion. To me rule is not clear about that. But I guess to be safe, I should just make a disclosure. No harm in doing that.\n\n> provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.\n\nthere is no \"disclosure rule\", merely a \"fairness clause\". This is how I'm interpreting it\n\n`Please use this thread to list the pre-trained models you are planning to use, or other external data. Once a model or data source has been posted once, it does not need to be posted again. You must post here one week prior to competition close.` -Inversion\n\nsince the external data I'm planning on using is already posted per the \"fairness clause\", I see no need to individual state that I am using it." }, { "id": 537228, "postDate": "2019-05-26T14:05:16.880Z", "content": "It is very clear: the source of data must be posted, not necessarily by you, which was your question, but you, and no one else, must have the right to use that data for the competition purpose. It means looking at whatever license, copyright, patent, whatever that can limit rights to use that data source.
", "rawMarkdown": "It is very clear: the source of data must be posted, not necessarily by you, which was your question, but you, and no one else, must have the right to use that data for the competition purpose. It means looking at whatever license, copyright, patent, whatever that can limit rights to use that data source." } ] }, { "id": 458602, "postDate": "2019-01-20T03:12:25.270Z", "content": "Hi
\n\nIs this competition a replacement for DSB2019, or will those continue?
\n\nBy the way, am I reading the Rules/Timeline correctly as implying that there will be no second stage here?
", "rawMarkdown": "Hi\n\nIs this competition a replacement for DSB2019, or will those continue?\n\nBy the way, am I reading the Rules/Timeline correctly as implying that there will be no second stage here? ", "replies": [ { "id": 458970, "postDate": "2019-01-20T23:22:41.657Z", "content": "Hi Oleg -
\n\nThis competition is independent of DSB.
\n\nAnd correct, there is no second stage.
", "rawMarkdown": "Hi Oleg -\n\nThis competition is independent of DSB.\n\nAnd correct, there is no second stage." } ] }, { "id": 454692, "postDate": "2019-01-12T02:54:52.013Z", "content": "I'll open the source code. But it will be in my chosen programming language. Is this okay? Will the decision count?
", "rawMarkdown": "I'll open the source code. But it will be in my chosen programming language. Is this okay? Will the decision count?", "replies": [ { "id": 454707, "postDate": "2019-01-12T04:03:33.100Z", "content": "You only have to open-source the code if you win. As long as the open-sourcing is in adherence to opensource.org, you can use whatever you want! Happy coding!
", "rawMarkdown": "You only have to open-source the code if you win. As long as the open-sourcing is in adherence to opensource.org, you can use whatever you want! Happy coding!", "votes": 3 } ] }, { "id": 464516, "postDate": "2019-02-01T02:44:33.857Z", "content": "Interesting challenge. would love to try.
", "rawMarkdown": "Interesting challenge. would love to try.", "isDeleted": true }, { "id": 454691, "postDate": "2019-01-12T02:50:48.733Z", "rawMarkdown": "", "votes": -3, "isDeleted": true }, { "id": 454231, "postDate": "2019-01-11T09:20:42.180Z", "content": "Sounds fun. Thank you for organizing.
", "rawMarkdown": "Sounds fun. Thank you for organizing.", "votes": 6 } ], "comments": [ { "id": 2142772, "author_name": "Manas Johri", "author_url": "", "post_date": "2023-02-13T19:03:26.573000", "content": "Hi. I'm confused about feature engineering. why do we take the last value of the target for every segment (150000).
", "votes": 1, "replies": [] }, { "id": 454546, "author_name": "Carlo Graziani", "author_url": "", "post_date": "2019-01-11T20:16:13.547000", "content": "Hi. This looks interesting.
\n\nI have a methodological, well, not \"quibble\", but suggestion for the future, at any rate: forecasts tend to be more valuable and actionable if they are probabilistic, rather than point-valued. That is, a full pdf of the time-to-failure (in your application) would be more useful for making decisions under uncertainty about the future than would a single summary statistic from the pdf, such as the mean or the mode. In weather forecasting (and in other applications, such as finance, epidemiology, etc.) this style of forecasting is common -- the forecast is a pdf conditioned on current information. Not only does this enable more powerful forecast assessment skill scores, but it also makes possible more powerful risk management methodologies.
", "votes": 14, "replies": [] }, { "id": 519203, "author_name": "Santhanalakshmi", "author_url": "", "post_date": "2019-04-18T14:37:26.280000", "content": "Hi I didn't understand what is meant by predicting the remaining time. Is it the remaining time between current earthquake and occurrence of next earthquake ? Also I'm new to Kaggle. Can I use python tensorflow for this implementation ?
", "votes": 7, "replies": [ { "id": 533955, "author_name": "SySu", "author_url": "", "post_date": "2019-05-20T10:22:43.133000", "content": "Yes, it the remaining time between the current earthquake and the occurrence of the next earthquake. And, you can either use R or Python.
", "votes": 0, "replies": [] } ] }, { "id": 454381, "author_name": "Chris460", "author_url": "", "post_date": "2019-01-11T14:58:14.020000", "content": "Is MATLAB acceptable for the competition? If not, would I need to use a free trial license to MATLAB issued to me for the express use of competing on this project?
", "votes": 1, "replies": [ { "id": 454481, "author_name": "Addison Howard", "author_url": "", "post_date": "2019-01-11T17:48:12.350000", "content": "Hi Chris,
\n\nPer the rules of the competition, all solutions must be open-sourced. As Matlab would require a paid license for the host to use, it would not be considered an open-sourced solutions.
\n\nThanks!
", "votes": 4, "replies": [] }, { "id": 455780, "author_name": "DevilEars", "author_url": "", "post_date": "2019-01-14T15:25:59.373000", "content": "As a curiosity, would Octave code be acceptable then?
", "votes": 4, "replies": [] } ] }, { "id": 456405, "author_name": "Vladimir Lukin", "author_url": "", "post_date": "2019-01-15T18:47:21.643000", "content": "I believe that the data must be 'chunked' by the earthquake events. Is it true that anything that happened at cycle k, (cycle is time between observations, where time-to-event=0), should not matter to then next cycle k+1? I wonder if test data sampled strictly from the 'before event' window.
", "votes": 2, "replies": [] }, { "id": 454749, "author_name": "Chris460", "author_url": "", "post_date": "2019-01-12T06:10:40.447000", "content": "Hello Addison.
\n\nMathworks (makers of MATLAB) has hosted Kaggle competitions before and as part of their sponsorship, they have provided free licenses to MATLAB and any necessary toolboxes for the duration of the competition. Their algorithms are not hidden - they are available for the user to peruse. Would this be an option for this competition? I spoke with a MATLAB representative this morning and he was interested in working something out so that myself and any other users could freely access MATLAB for the purpose of the competition. BTW, I have seen Q&As with former Kaggle personnel where MATLAB was allowed. Are you absolutely certain using MATLAB is banned from being an acceptable entry for a winning solution?
\n\nHere is an example where MATLAB was used: https://www.kaggle.com/c/melbourne-university-seizure-prediction
\n\nBest regards\nChris
", "votes": 2, "replies": [ { "id": 454752, "author_name": "Chris460", "author_url": "", "post_date": "2019-01-12T06:16:22.870000", "content": "More info about where MATLAB was an acceptable for a Kaggle competition: https://www.kaggle.com/c/noaa-right-whale-recognition/discussion/16161#91048
", "votes": 0, "replies": [] }, { "id": 455132, "author_name": "XinLi", "author_url": "", "post_date": "2019-01-13T03:20:37.747000", "content": "Hi Chris, Thank you for your information. I am interested to use Matlab too. Please let me know Mathworks decision. I have used Matlab since early 90s and bought commercial license myself 10 years ago - using matlab would make things easier if the NN toolbox is included. The kaggle users can really do them a favor to test the toolbox. I am seriously considering to move to Python because of progress of Jupyter Notebook. Again thanks.
", "votes": 0, "replies": [] } ] }, { "id": 1348762, "author_name": "Ольга Гусева", "author_url": "", "post_date": "2021-06-14T08:25:59.623000", "content": "Hello! Could you please tell me why in the trainig data we have time_to_failure for each acoustic signal, but in the result we have one value of time_to_failure for 150 000 values of acoustic data? How can I compare my result with what is given?
", "votes": 0, "replies": [] }, { "id": 710205, "author_name": "Petro Shopin", "author_url": "", "post_date": "2020-01-04T12:59:08.547000", "content": "That was a great competition!
", "votes": 0, "replies": [] }, { "id": 558277, "author_name": "Raghvendra Pal", "author_url": "", "post_date": "2019-06-22T04:38:27.457000", "content": "Hi, Today i have submitted my result file of LANLEarthquake Prediction, But I am not getting my name in Leaderboard,\nI do not understand why? Can anyone tell me?
", "votes": 0, "replies": [] }, { "id": 531627, "author_name": "Vishal Sharma", "author_url": "", "post_date": "2019-05-15T08:22:53.470000", "content": "nice.
", "votes": 0, "replies": [] }, { "id": 531623, "author_name": "林湧森 (Dyson Lin)", "author_url": "", "post_date": "2019-05-15T08:08:16.120000", "content": "Help! I could not load the LANL data set in my kernel.\nCould the Kaggle Team check the dataset?\nThanks!
", "votes": 0, "replies": [] }, { "id": 519186, "author_name": "Fatih Öztürk", "author_url": "", "post_date": "2019-04-18T14:14:32.577000", "content": "Anyone else facing 404 - Not Found problem while downloading datasets via kaggle api command ?\nEDIT: Just checked other competitions' kaggle api command and they are giving same error as well.
just in case you hadn't come across the solution and it's the same thing I was facing on colab, changing over to a CPU kernel to download the data did the trick
", "votes": 0, "replies": [] } ] }, { "id": 492377, "author_name": "Sourabh", "author_url": "", "post_date": "2019-03-17T07:00:35.873000", "content": "Interesting problem statement, especially for those who enjoy physics. I have used Svm SVR for this, and got 1.69 score . How can I improve my model, any suggestions?
", "votes": 0, "replies": [] }, { "id": 480609, "author_name": "Stanislav Blinov", "author_url": "", "post_date": "2019-02-28T12:15:22.663000", "content": "Hi!\nCould you please clarify, what does that mean: \"External Data Disclosure deadline. All external data used in the competition must be disclosed in the forums by this date.\" - What is \"external data\" in this particular case?\nAlso, am I right assuming that there'll be no private leaderboard and no additional data for final scoring?
\n\np.s. sorry for dumb questions, this competition is baptism of fire on kaggle :)\nThanks!
", "votes": 0, "replies": [ { "id": 480808, "author_name": "Addison Howard", "author_url": "", "post_date": "2019-02-28T17:36:42.197000", "content": "Hi Stanislav,
\n\nIf you use any data or pre-trained models not provided on the competition Data page, you must disclose what they are before the competition data deadline.
\n\nThere is a private leaderboard, but it is a subset of the full test set.
", "votes": 2, "replies": [] }, { "id": 480817, "author_name": "Stanislav Blinov", "author_url": "", "post_date": "2019-02-28T17:58:28.543000", "content": "Thanks for your reply!
", "votes": 0, "replies": [] }, { "id": 536918, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-25T16:29:13.870000", "content": "@addisonhoward @inversion Can we use external data or pre-trained model that has been disclosed/mentioned by other people during the competition, or is it an individual disclosure letting the kaggle team and others you are using external data/pre-trained models. Hoping for last minute clarification. Thanks!
", "votes": 0, "replies": [] }, { "id": 537099, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-26T07:13:51.843000", "content": "@teeyee314 This is covered in the competition rules, external data section. You have to disclose what you use in the competition forum at least one week before competition end:
\n\n\n", "votes": 1, "replies": [] }, { "id": 537111, "author_name": "Tim Yee", "author_url": "", "post_date": "2019-05-26T08:01:02.550000", "content": "EXTERNAL DATA
\n \nThe following provision supersedes General Rules Section 7.C. below: “You may use data, other than the Competition Data, as allowed on the Competition Website to develop and test your models and Submissions; provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.\"
\n
@cpmpml I read that thoroughly. It is not clear if I need publicly say that I am using the external data shared by other people in the discussion. To me rule is not clear about that. But I guess to be safe, I should just make a disclosure. No harm in doing that.
\n\n> provided, you have the right and authority to use such external data for the purposes of the Competition, and to share such data with Sponsor and Kaggle as may be required. The source of any external data must be posted to the official competition forum prior to the Entry Deadline.
\n\nthere is no \"disclosure rule\", merely a \"fairness clause\". This is how I'm interpreting it
\n\nPlease use this thread to list the pre-trained models you are planning to use, or other external data. Once a model or data source has been posted once, it does not need to be posted again. You must post here one week prior to competition close. -Inversion
since the external data I'm planning on using is already posted per the \"fairness clause\", I see no need to individual state that I am using it.
", "votes": 0, "replies": [] }, { "id": 537228, "author_name": "CPMP", "author_url": "", "post_date": "2019-05-26T14:05:16.880000", "content": "It is very clear: the source of data must be posted, not necessarily by you, which was your question, but you, and no one else, must have the right to use that data for the competition purpose. It means looking at whatever license, copyright, patent, whatever that can limit rights to use that data source.
", "votes": 0, "replies": [] } ] }, { "id": 458602, "author_name": "Oleg Trott", "author_url": "", "post_date": "2019-01-20T03:12:25.270000", "content": "Hi
\n\nIs this competition a replacement for DSB2019, or will those continue?
\n\nBy the way, am I reading the Rules/Timeline correctly as implying that there will be no second stage here?
", "votes": 0, "replies": [ { "id": 458970, "author_name": "inversion", "author_url": "", "post_date": "2019-01-20T23:22:41.657000", "content": "Hi Oleg -
\n\nThis competition is independent of DSB.
\n\nAnd correct, there is no second stage.
", "votes": 0, "replies": [] } ] }, { "id": 454692, "author_name": "Eugene Lutsenko", "author_url": "", "post_date": "2019-01-12T02:54:52.013000", "content": "I'll open the source code. But it will be in my chosen programming language. Is this okay? Will the decision count?
", "votes": 0, "replies": [ { "id": 454707, "author_name": "Addison Howard", "author_url": "", "post_date": "2019-01-12T04:03:33.100000", "content": "You only have to open-source the code if you win. As long as the open-sourcing is in adherence to opensource.org, you can use whatever you want! Happy coding!
", "votes": 3, "replies": [] } ] }, { "id": 464516, "author_name": "", "author_url": "", "post_date": "2019-02-01T02:44:33.857000", "content": "Interesting challenge. would love to try.
", "votes": 0, "replies": [] }, { "id": 454691, "author_name": "", "author_url": "", "post_date": "2019-01-12T02:50:48.733000", "content": "", "votes": -3, "replies": [] }, { "id": 454231, "author_name": "Dieter", "author_url": "", "post_date": "2019-01-11T09:20:42.180000", "content": "Sounds fun. Thank you for organizing.
", "votes": 6, "replies": [] } ], "raw_markdown_by_id": { "2142772": "Hi. I'm confused about feature engineering. why do we take the last value of the target for every segment (150000).", "453856": "The goal of this competition is to use seismic signals to predict the timing of laboratory earthquakes. The data comes from a well-known experimental set-up used to study earthquake physics. Can the Kaggle community can help push the boundaries of this scientific research?\n\nIf you want to get started fast, this [Basic Feature Benchmark Kernel](https://www.kaggle.com/inversion/basic-feature-benchmark) might be helpful.\n\nPlease feel free to ask your questions in this thread.\n\nGood luck!", "454546": "Hi. This looks interesting.\n\nI have a methodological, well, not \"quibble\", but suggestion for the future, at any rate: forecasts tend to be more valuable and actionable if they are probabilistic, rather than point-valued. That is, a full pdf of the time-to-failure (in your application) would be more useful for making decisions under uncertainty about the future than would a single summary statistic from the pdf, such as the mean or the mode. In weather forecasting (and in other applications, such as finance, epidemiology, etc.) this style of forecasting is common -- the forecast is a pdf conditioned on current information. Not only does this enable more powerful forecast assessment skill scores, but it also makes possible more powerful risk management methodologies.", "519203": "Hi I didn't understand what is meant by predicting the remaining time. Is it the remaining time between current earthquake and occurrence of next earthquake ? Also I'm new to Kaggle. Can I use python tensorflow for this implementation ?", "454381": "Is MATLAB acceptable for the competition? If not, would I need to use a free trial license to MATLAB issued to me for the express use of competing on this project?", "456405": "I believe that the data must be 'chunked' by the earthquake events. Is it true that anything that happened at cycle k, (cycle is time between observations, where time-to-event=0), should not matter to then next cycle k+1? I wonder if test data sampled strictly from the 'before event' window. ", "454749": "Hello Addison.\n\nMathworks (makers of MATLAB) has hosted Kaggle competitions before and as part of their sponsorship, they have provided free licenses to MATLAB and any necessary toolboxes for the duration of the competition. Their algorithms are not hidden - they are available for the user to peruse. Would this be an option for this competition? I spoke with a MATLAB representative this morning and he was interested in working something out so that myself and any other users could freely access MATLAB for the purpose of the competition. BTW, I have seen Q&As with former Kaggle personnel where MATLAB was allowed. Are you absolutely certain using MATLAB is banned from being an acceptable entry for a winning solution?\n\nHere is an example where MATLAB was used: https://www.kaggle.com/c/melbourne-university-seizure-prediction\n\nBest regards\nChris", "1348762": "Hello! Could you please tell me why in the trainig data we have time_to_failure for each acoustic signal, but in the result we have one value of time_to_failure for 150 000 values of acoustic data? How can I compare my result with what is given?", "710205": "That was a great competition! ", "558277": "Hi, Today i have submitted my result file of LANLEarthquake Prediction, But I am not getting my name in Leaderboard,\nI do not understand why? Can anyone tell me?", "531627": "nice.", "531623": "Help! I could not load the LANL data set in my kernel.\nCould the Kaggle Team check the dataset?\nThanks!", "519186": "Anyone else facing `404 - Not Found` problem while downloading datasets via kaggle api command ?\nEDIT: Just checked other competitions' kaggle api command and they are giving same error as well.", "492377": "Interesting problem statement, especially for those who enjoy physics. I have used Svm SVR for this, and got 1.69 score . How can I improve my model, any suggestions?", "480609": "Hi!\nCould you please clarify, what does that mean: \"External Data Disclosure deadline. All external data used in the competition must be disclosed in the forums by this date.\" - What is \"external data\" in this particular case?\nAlso, am I right assuming that there'll be no private leaderboard and no additional data for final scoring?\n\np.s. sorry for dumb questions, this competition is baptism of fire on kaggle :)\nThanks!", "458602": "Hi\n\nIs this competition a replacement for DSB2019, or will those continue?\n\nBy the way, am I reading the Rules/Timeline correctly as implying that there will be no second stage here? ", "454692": "I'll open the source code. But it will be in my chosen programming language. Is this okay? Will the decision count?", "464516": "Interesting challenge. would love to try.", "454691": "", "454231": "Sounds fun. Thank you for organizing." } }