Welcome to the TrackML Particle Tracking Challenge, hosted by CERN.
\n\nIn this competition, you’re challenged with building an algorithm that accurately reconstructs particle tracks from 3D points left in silicon detectors.
\n\nWe're happy to announce that this has been launched as a Featured competition, meaning prizes, points, and medals!
\n\nThe Training data is >200 Gb uncompressed. We've split it into 5 parts to make it easier to download. We've also included a 100-event \"train_sample\" if you want to explore the data before deciding to download the entire set.
\n\nNote: Due to the number and size of the data files, we've only included the files from \"train_1\" in the Kernels environment (along with the full Test set, etc.)
\n\nGood luck!
", "messages": [ { "id": 321223, "postDate": "2018-04-30T21:07:16.450Z", "content": "Welcome to the TrackML Particle Tracking Challenge, hosted by CERN.
\n\nIn this competition, you’re challenged with building an algorithm that accurately reconstructs particle tracks from 3D points left in silicon detectors.
\n\nWe're happy to announce that this has been launched as a Featured competition, meaning prizes, points, and medals!
\n\nThe Training data is >200 Gb uncompressed. We've split it into 5 parts to make it easier to download. We've also included a 100-event \"train_sample\" if you want to explore the data before deciding to download the entire set.
\n\nNote: Due to the number and size of the data files, we've only included the files from \"train_1\" in the Kernels environment (along with the full Test set, etc.)
\n\nGood luck!
", "rawMarkdown": "Welcome to the TrackML Particle Tracking Challenge, hosted by CERN.\n\nIn this competition, you’re challenged with building an algorithm that accurately reconstructs particle tracks from 3D points left in silicon detectors.\n\nWe're happy to announce that this has been launched as a Featured competition, meaning prizes, points, and medals!\n\nThe Training data is >200 Gb uncompressed. We've split it into 5 parts to make it easier to download. We've also included a 100-event \"train_sample\" if you want to explore the data before deciding to download the entire set.\n\n**Note:** Due to the number and size of the data files, we've only included the files from \"train_1\" in the Kernels environment (along with the full Test set, etc.)\n\nGood luck!", "votes": 29 }, { "id": 321525, "postDate": "2018-05-01T13:34:07.277Z", "content": "Thank you to the organizers for setting up the competition. The data acquisition library is awesome and makes slicing a lot easier. From first glance having an understanding about the physics might give a little edge. Do you all think teams with more physicists will craft a better model, or will it come down to expertise in data science/ machine learning that cracks the tracks? Best luck to everyone!
", "rawMarkdown": "Thank you to the organizers for setting up the competition. The data acquisition library is awesome and makes slicing a lot easier. From first glance having an understanding about the physics might give a little edge. Do you all think teams with more physicists will craft a better model, or will it come down to expertise in data science/ machine learning that cracks the tracks? Best luck to everyone!", "votes": 3, "replies": [ { "id": 321531, "postDate": "2018-05-01T13:40:29.843Z", "content": "The physics effects we have included in the simulation are relatively mild, compared to the combinatorial problem. At first glance one could assume all particles to progress along a helical trajectory - which is the movement of a charged particle in constant magnetic field.
", "rawMarkdown": "The physics effects we have included in the simulation are relatively mild, compared to the combinatorial problem. At first glance one could assume all particles to progress along a helical trajectory - which is the movement of a charged particle in constant magnetic field.", "votes": 6 }, { "id": 321801, "postDate": "2018-05-01T23:21:47.140Z", "content": "are you getting rid of loopers as well?
", "rawMarkdown": "are you getting rid of loopers as well?", "votes": 1 }, { "id": 321880, "postDate": "2018-05-02T04:41:18.113Z", "content": "We have run the simulation with a minimum momentum requirement, which practically does not allow loopers.
", "rawMarkdown": "We have run the simulation with a minimum momentum requirement, which practically does not allow loopers. ", "votes": 2 }, { "id": 322047, "postDate": "2018-05-02T11:00:31.890Z", "content": "Great! One other question would be on the global fitting of the tracks. Could you elaborate on how this is performed? I have worked with tracking in other HEP experiments in the past and was wondering about the DCA from the vertex (how is it calculated?). Finally you mention simulation, therefore we are talking about geant4 tracks?
", "rawMarkdown": "Great! One other question would be on the global fitting of the tracks. Could you elaborate on how this is performed? I have worked with tracking in other HEP experiments in the past and was wondering about the DCA from the vertex (how is it calculated?). Finally you mention simulation, therefore we are talking about geant4 tracks?\n", "votes": 1 }, { "id": 322053, "postDate": "2018-05-02T11:11:44.447Z", "content": "We deliberately do not ask for the fitted parameters, hence track fitting is not part of the challenge. We are only interested in the grouping of hits. This is done using a fast simulation engine (not Geant4), in order to be able to simplify the data a bit.
", "rawMarkdown": "We deliberately do not ask for the fitted parameters, hence track fitting is not part of the challenge. We are only interested in the *grouping* of hits. This is done using a fast simulation engine (not Geant4), in order to be able to simplify the data a bit. ", "votes": 3 }, { "id": 340471, "postDate": "2018-06-09T09:04:26.990Z", "content": "What are \"loopers\"? Are you talking about recursive neural networks?
", "rawMarkdown": "What are \"loopers\"? Are you talking about recursive neural networks?" }, { "id": 340555, "postDate": "2018-06-09T15:27:23.830Z", "content": "Low pt particles That are following a circular path inside the detectors
", "rawMarkdown": "Low pt particles That are following a circular path inside the detectors" } ] }, { "id": 321268, "postDate": "2018-04-30T22:30:38.537Z", "content": "Hi organizers.\nThank you for arranging such a fascinating challenge!\nIt would be great if you could provide some graph or sketch of the detector which could help to understand a construction better. It would be helpful to relate certain features to the part of the detector. Thank you in advance.
", "rawMarkdown": "Hi organizers.\nThank you for arranging such a fascinating challenge!\nIt would be great if you could provide some graph or sketch of the detector which could help to understand a construction better. It would be helpful to relate certain features to the part of the detector. Thank you in advance.", "votes": 3, "replies": [ { "id": 321366, "postDate": "2018-05-01T05:22:39.607Z", "content": "There will be a full detector and dataset description document published today including all the details, apologies for the asynchronous start.
", "rawMarkdown": "There will be a full detector and dataset description document published today including all the details, apologies for the asynchronous start.", "votes": 4 }, { "id": 321529, "postDate": "2018-05-01T13:37:18.483Z", "content": "Here's a beta version of the detector description document:\nhttps://cernbox.cern.ch/index.php/s/tkF9YFCrYu8auWw
\n\nAnd the 3D event display:\nhttps://emoyse.web.cern.ch/emoyse/WebEventDisplay/jsdisplay_TrackML.html
", "rawMarkdown": "Here's a beta version of the detector description document:\nhttps://cernbox.cern.ch/index.php/s/tkF9YFCrYu8auWw\n\nAnd the 3D event display:\nhttps://emoyse.web.cern.ch/emoyse/WebEventDisplay/jsdisplay_TrackML.html", "votes": 3 } ] }, { "id": 321293, "postDate": "2018-05-01T00:19:19.030Z", "content": "So this is the “big” dataset that we’ve been promised? I’m disappointed. It’s got nothing on TSA. :P
", "rawMarkdown": "So this is the “big” dataset that we’ve been promised? I’m disappointed. It’s got nothing on TSA. :P", "votes": 4, "replies": [ { "id": 321300, "postDate": "2018-05-01T00:54:28.963Z", "content": "LOL. No, the BIG one is still coming! ;-)
", "rawMarkdown": "LOL. No, the BIG one is still coming! ;-)", "votes": 10 }, { "id": 321581, "postDate": "2018-05-01T15:48:41.717Z", "content": "TSA 2:
\n\nInvasion of the Body Scanners
\n\nA Boddy Scanner Darkly
", "rawMarkdown": "TSA 2: \n\nInvasion of the Body Scanners\n\nA Boddy Scanner Darkly " } ] }, { "id": 321275, "postDate": "2018-04-30T23:02:47.333Z", "content": "Maybe these top 3 solutions from some inclass competition on tracks from OPERA experiment will be helpful for start.\nhttps://www.kaggle.com/c/dark-matter-signal-search-episode-1/discussion/36725
", "rawMarkdown": "Maybe these top 3 solutions from some inclass competition on tracks from OPERA experiment will be helpful for start.\nhttps://www.kaggle.com/c/dark-matter-signal-search-episode-1/discussion/36725", "votes": 4 }, { "id": 322379, "postDate": "2018-05-02T20:36:36.327Z", "content": "This is very interesting challenge but I think we need to know physics alot....
", "rawMarkdown": "This is very interesting challenge but I think we need to know physics alot....", "votes": 1, "replies": [ { "id": 322392, "postDate": "2018-05-02T21:07:50.360Z", "content": "you got a strong point but judging by winners in previous physics (kaggle) contest, they weren't physicists at all.
", "rawMarkdown": "you got a strong point but judging by winners in previous physics (kaggle) contest, they weren't physicists at all.", "votes": 2 }, { "id": 322413, "postDate": "2018-05-02T22:15:41.397Z", "content": "Actually it is really about discrete optimisation in 3D, connecting together points with arc of circles (well arcs of helices in 3D), there is no sophisticated physics.
", "rawMarkdown": "Actually it is really about discrete optimisation in 3D, connecting together points with arc of circles (well arcs of helices in 3D), there is no sophisticated physics.", "votes": 3 }, { "id": 322414, "postDate": "2018-05-02T22:18:40.340Z", "content": "I'm a Physicist, but don't expect to use much (if any) of my actual Physics knowledge here.
", "rawMarkdown": "I'm a Physicist, but don't expect to use much (if any) of my actual Physics knowledge here. ", "votes": 2 } ] }, { "id": 321271, "postDate": "2018-04-30T22:45:39.957Z", "content": "I missed out on the previous CERN competitions, glad I am not missing this one!
", "rawMarkdown": "I missed out on the previous CERN competitions, glad I am not missing this one!", "votes": 2 }, { "id": 374785, "postDate": "2018-08-23T20:00:50.320Z", "content": "The TrackML Kaggle competition has ended. Any participant (not necessarily with the highest score) who think they made valuable contributions with innovative algorithms (in particular if they have shared insights on the forum) are welcome to release publicly their code with an open source license and lightweight structured documentation. This will allow to compete for the « HEP meets ML » jury prizes: a NVidia V100 GPU, and two invitations to NIPS 2018 or the spring 2019 CERN workshop.See details in forum topic:
\n\nhttps://www.kaggle.com/c/trackml-particle-identification/discussion/63289
", "rawMarkdown": "The TrackML Kaggle competition has ended. Any participant (not necessarily with the highest score) who think they made valuable contributions with innovative algorithms (in particular if they have shared insights on the forum) are welcome to release publicly their code with an open source license and lightweight structured documentation. This will allow to compete for the « HEP meets ML » jury prizes: a NVidia V100 GPU, and two invitations to NIPS 2018 or the spring 2019 CERN workshop.See details in forum topic:\n \nhttps://www.kaggle.com/c/trackml-particle-identification/discussion/63289" }, { "id": 353645, "postDate": "2018-07-07T10:16:28.223Z", "content": "hi David I have a track_id attached to this message is ggplot of this. I was wondering is the tracker supposed to unify micro and macro events, because the only way I was able to make a positive tracker was using something like the Legendrian so the tracker is not north of strangeness and spontaneous. Does this come close to what you need? How do I complete my work if I do not have access to training sets 2-5? I called the plot something like tracking the destiny of future language because the \"macro\" events tether a Morse like line. I also think the curve looks like the two body Coulomb attraction/repulsion curve over distance. help me out if it is worth it and tell me what you think.
", "rawMarkdown": "hi David I have a track_id attached to this message is ggplot of this. I was wondering is the tracker supposed to unify micro and macro events, because the only way I was able to make a positive tracker was using something like the Legendrian so the tracker is not north of strangeness and spontaneous. Does this come close to what you need? How do I complete my work if I do not have access to training sets 2-5? I called the plot something like tracking the destiny of future language because the \"macro\" events tether a Morse like line. I also think the curve looks like the two body Coulomb attraction/repulsion curve over distance. help me out if it is worth it and tell me what you think." }, { "id": 322668, "postDate": "2018-05-03T11:57:21.297Z", "content": "Is that convinent with the very beginners?
", "rawMarkdown": "Is that convinent with the very beginners?" }, { "id": 321485, "postDate": "2018-05-01T12:01:29.310Z", "content": "Looks interesting.
", "rawMarkdown": "Looks interesting." }, { "id": 321335, "postDate": "2018-05-01T04:08:46.040Z", "rawMarkdown": "", "votes": 1, "isDeleted": true, "replies": [ { "id": 321476, "postDate": "2018-05-01T11:38:42.127Z", "content": "Facing the same issue on basic submission. As per the data page, submission should have 3 columns and not 5 as hinted by the error msg above.
\n\n3rd column should be the track_id which should be a self choosen/user-defined identifier (positive int) of the track id so I don't understand that error message.
\n\nAlso sample submission file has 13741467 lines but submission form states : \"We expect the solution file to have 3343036 prediction rows\" ?!?
\n\nLooks like an error in scorer or sample submission provided.
", "rawMarkdown": "Facing the same issue on basic submission. As per the data page, submission should have 3 columns and not 5 as hinted by the error msg above.\n\n3rd column should be the track_id which should be a self choosen/user-defined identifier (positive int) of the track id so I don't understand that error message.\n\nAlso sample submission file has 13741467 lines but submission form states : \"We expect the solution file to have 3343036 prediction rows\" ?!?\n\nLooks like an error in scorer or sample submission provided.", "votes": 1 }, { "id": 321497, "postDate": "2018-05-01T12:30:44.227Z", "content": "A clarification: the input dataset that you've received contains 3d coordinates for the locations where the particles passed through the detector (hits), however, the rows of the submission should contain event_id, hit_id and track_id from the tracks that you've found, not 3d position coordinates. The hits are labelled by a hit_id. See here: https://www.kaggle.com/c/trackml-particle-identification#evaluation
", "rawMarkdown": "A clarification: the input dataset that you've received contains 3d coordinates for the locations where the particles passed through the detector (hits), however, the rows of the submission should contain event_id, hit_id and track_id from the tracks that you've found, not 3d position coordinates. The hits are labelled by a hit_id. See here: https://www.kaggle.com/c/trackml-particle-identification#evaluation" }, { "id": 321513, "postDate": "2018-05-01T13:18:28.853Z", "content": "exactly what I've done :
\n\nhead -3 submission1.csv \nevent_id,hit_id,track_id\n0,1,1\n0,2,1\nand still getting
\n\nERROR: The value '1' is not in the required set of values 'Non-negative = [0, ∞)' (Line 2, Column 5)\nApologies . . . it appears there was a bug in our validation code. This has been resolved.
", "rawMarkdown": "Apologies . . . it appears there was a bug in our validation code. This has been resolved.", "votes": 1 }, { "id": 335264, "postDate": "2018-05-29T13:13:15.707Z", "content": "So is there a 3343036 prediction rows restriction after all? (the total number of hits in the test dataset is 13741466)
", "rawMarkdown": "So is there a 3343036 prediction rows restriction after all? (the total number of hits in the test dataset is 13741466) " }, { "id": 336961, "postDate": "2018-06-01T16:15:25.043Z", "content": "The correct number is the same as what's in the sample_submission.csv and test data set: 13741466.
Thank you for the prompt reply!
", "rawMarkdown": "Thank you for the prompt reply!" } ] } ], "comments": [ { "id": 321525, "author_name": "AlanVitullo", "author_url": "", "post_date": "2018-05-01T13:34:07.277000", "content": "Thank you to the organizers for setting up the competition. The data acquisition library is awesome and makes slicing a lot easier. From first glance having an understanding about the physics might give a little edge. Do you all think teams with more physicists will craft a better model, or will it come down to expertise in data science/ machine learning that cracks the tracks? Best luck to everyone!
", "votes": 3, "replies": [ { "id": 321531, "author_name": "Andreas Salzburger", "author_url": "", "post_date": "2018-05-01T13:40:29.843000", "content": "The physics effects we have included in the simulation are relatively mild, compared to the combinatorial problem. At first glance one could assume all particles to progress along a helical trajectory - which is the movement of a charged particle in constant magnetic field.
", "votes": 6, "replies": [] }, { "id": 321801, "author_name": "Art", "author_url": "", "post_date": "2018-05-01T23:21:47.140000", "content": "are you getting rid of loopers as well?
", "votes": 1, "replies": [] }, { "id": 321880, "author_name": "Andreas Salzburger", "author_url": "", "post_date": "2018-05-02T04:41:18.113000", "content": "We have run the simulation with a minimum momentum requirement, which practically does not allow loopers.
", "votes": 2, "replies": [] }, { "id": 322047, "author_name": "Art", "author_url": "", "post_date": "2018-05-02T11:00:31.890000", "content": "Great! One other question would be on the global fitting of the tracks. Could you elaborate on how this is performed? I have worked with tracking in other HEP experiments in the past and was wondering about the DCA from the vertex (how is it calculated?). Finally you mention simulation, therefore we are talking about geant4 tracks?
", "votes": 1, "replies": [] }, { "id": 322053, "author_name": "Andreas Salzburger", "author_url": "", "post_date": "2018-05-02T11:11:44.447000", "content": "We deliberately do not ask for the fitted parameters, hence track fitting is not part of the challenge. We are only interested in the grouping of hits. This is done using a fast simulation engine (not Geant4), in order to be able to simplify the data a bit.
", "votes": 3, "replies": [] }, { "id": 340471, "author_name": "Corey Carlson", "author_url": "", "post_date": "2018-06-09T09:04:26.990000", "content": "What are \"loopers\"? Are you talking about recursive neural networks?
", "votes": 0, "replies": [] }, { "id": 340555, "author_name": "Art", "author_url": "", "post_date": "2018-06-09T15:27:23.830000", "content": "Low pt particles That are following a circular path inside the detectors
", "votes": 0, "replies": [] } ] }, { "id": 321268, "author_name": "Lukasz Aszyk", "author_url": "", "post_date": "2018-04-30T22:30:38.537000", "content": "Hi organizers.\nThank you for arranging such a fascinating challenge!\nIt would be great if you could provide some graph or sketch of the detector which could help to understand a construction better. It would be helpful to relate certain features to the part of the detector. Thank you in advance.
", "votes": 3, "replies": [ { "id": 321366, "author_name": "Andreas Salzburger", "author_url": "", "post_date": "2018-05-01T05:22:39.607000", "content": "There will be a full detector and dataset description document published today including all the details, apologies for the asynchronous start.
", "votes": 4, "replies": [] }, { "id": 321529, "author_name": "Andreas Salzburger", "author_url": "", "post_date": "2018-05-01T13:37:18.483000", "content": "Here's a beta version of the detector description document:\nhttps://cernbox.cern.ch/index.php/s/tkF9YFCrYu8auWw
\n\nAnd the 3D event display:\nhttps://emoyse.web.cern.ch/emoyse/WebEventDisplay/jsdisplay_TrackML.html
", "votes": 3, "replies": [] } ] }, { "id": 321293, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-01T00:19:19.030000", "content": "So this is the “big” dataset that we’ve been promised? I’m disappointed. It’s got nothing on TSA. :P
", "votes": 4, "replies": [ { "id": 321300, "author_name": "inversion", "author_url": "", "post_date": "2018-05-01T00:54:28.963000", "content": "LOL. No, the BIG one is still coming! ;-)
", "votes": 10, "replies": [] }, { "id": 321581, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-01T15:48:41.717000", "content": "TSA 2:
\n\nInvasion of the Body Scanners
\n\nA Boddy Scanner Darkly
", "votes": 0, "replies": [] } ] }, { "id": 321275, "author_name": "Leonid Sinev", "author_url": "", "post_date": "2018-04-30T23:02:47.333000", "content": "Maybe these top 3 solutions from some inclass competition on tracks from OPERA experiment will be helpful for start.\nhttps://www.kaggle.com/c/dark-matter-signal-search-episode-1/discussion/36725
", "votes": 4, "replies": [] }, { "id": 322379, "author_name": "", "author_url": "", "post_date": "2018-05-02T20:36:36.327000", "content": "This is very interesting challenge but I think we need to know physics alot....
", "votes": 1, "replies": [ { "id": 322392, "author_name": "Art", "author_url": "", "post_date": "2018-05-02T21:07:50.360000", "content": "you got a strong point but judging by winners in previous physics (kaggle) contest, they weren't physicists at all.
", "votes": 2, "replies": [] }, { "id": 322413, "author_name": "David Rousseau", "author_url": "", "post_date": "2018-05-02T22:15:41.397000", "content": "Actually it is really about discrete optimisation in 3D, connecting together points with arc of circles (well arcs of helices in 3D), there is no sophisticated physics.
", "votes": 3, "replies": [] }, { "id": 322414, "author_name": "Bojan Tunguz", "author_url": "", "post_date": "2018-05-02T22:18:40.340000", "content": "I'm a Physicist, but don't expect to use much (if any) of my actual Physics knowledge here.
", "votes": 2, "replies": [] } ] }, { "id": 321271, "author_name": "giim", "author_url": "", "post_date": "2018-04-30T22:45:39.957000", "content": "I missed out on the previous CERN competitions, glad I am not missing this one!
", "votes": 2, "replies": [] }, { "id": 374785, "author_name": "JR", "author_url": "", "post_date": "2018-08-23T20:00:50.320000", "content": "The TrackML Kaggle competition has ended. Any participant (not necessarily with the highest score) who think they made valuable contributions with innovative algorithms (in particular if they have shared insights on the forum) are welcome to release publicly their code with an open source license and lightweight structured documentation. This will allow to compete for the « HEP meets ML » jury prizes: a NVidia V100 GPU, and two invitations to NIPS 2018 or the spring 2019 CERN workshop.See details in forum topic:
\n\nhttps://www.kaggle.com/c/trackml-particle-identification/discussion/63289
", "votes": 0, "replies": [] }, { "id": 353645, "author_name": "Aimé T Shangula", "author_url": "", "post_date": "2018-07-07T10:16:28.223000", "content": "hi David I have a track_id attached to this message is ggplot of this. I was wondering is the tracker supposed to unify micro and macro events, because the only way I was able to make a positive tracker was using something like the Legendrian so the tracker is not north of strangeness and spontaneous. Does this come close to what you need? How do I complete my work if I do not have access to training sets 2-5? I called the plot something like tracking the destiny of future language because the \"macro\" events tether a Morse like line. I also think the curve looks like the two body Coulomb attraction/repulsion curve over distance. help me out if it is worth it and tell me what you think.
", "votes": 0, "replies": [] }, { "id": 322668, "author_name": "Ahmed Abulkhair", "author_url": "", "post_date": "2018-05-03T11:57:21.297000", "content": "Is that convinent with the very beginners?
", "votes": 0, "replies": [] }, { "id": 321485, "author_name": "OttoC", "author_url": "", "post_date": "2018-05-01T12:01:29.310000", "content": "Looks interesting.
", "votes": 0, "replies": [] }, { "id": 321335, "author_name": "", "author_url": "", "post_date": "2018-05-01T04:08:46.040000", "content": "", "votes": 1, "replies": [ { "id": 321476, "author_name": "TeraFlops", "author_url": "", "post_date": "2018-05-01T11:38:42.127000", "content": "Facing the same issue on basic submission. As per the data page, submission should have 3 columns and not 5 as hinted by the error msg above.
\n\n3rd column should be the track_id which should be a self choosen/user-defined identifier (positive int) of the track id so I don't understand that error message.
\n\nAlso sample submission file has 13741467 lines but submission form states : \"We expect the solution file to have 3343036 prediction rows\" ?!?
\n\nLooks like an error in scorer or sample submission provided.
", "votes": 1, "replies": [] }, { "id": 321497, "author_name": "Heather Gray", "author_url": "", "post_date": "2018-05-01T12:30:44.227000", "content": "A clarification: the input dataset that you've received contains 3d coordinates for the locations where the particles passed through the detector (hits), however, the rows of the submission should contain event_id, hit_id and track_id from the tracks that you've found, not 3d position coordinates. The hits are labelled by a hit_id. See here: https://www.kaggle.com/c/trackml-particle-identification#evaluation
", "votes": 0, "replies": [] }, { "id": 321513, "author_name": "TeraFlops", "author_url": "", "post_date": "2018-05-01T13:18:28.853000", "content": "exactly what I've done :
\n\nhead -3 submission1.csv \nevent_id,hit_id,track_id\n0,1,1\n0,2,1\nand still getting
\n\nERROR: The value '1' is not in the required set of values 'Non-negative = [0, ∞)' (Line 2, Column 5)\nApologies . . . it appears there was a bug in our validation code. This has been resolved.
", "votes": 1, "replies": [] }, { "id": 335264, "author_name": "Art", "author_url": "", "post_date": "2018-05-29T13:13:15.707000", "content": "So is there a 3343036 prediction rows restriction after all? (the total number of hits in the test dataset is 13741466)
", "votes": 0, "replies": [] }, { "id": 336961, "author_name": "inversion", "author_url": "", "post_date": "2018-06-01T16:15:25.043000", "content": "The correct number is the same as what's in the sample_submission.csv and test data set: 13741466.
Thank you for the prompt reply!
", "votes": 0, "replies": [] } ] } ], "raw_markdown_by_id": { "321223": "Welcome to the TrackML Particle Tracking Challenge, hosted by CERN.\n\nIn this competition, you’re challenged with building an algorithm that accurately reconstructs particle tracks from 3D points left in silicon detectors.\n\nWe're happy to announce that this has been launched as a Featured competition, meaning prizes, points, and medals!\n\nThe Training data is >200 Gb uncompressed. We've split it into 5 parts to make it easier to download. We've also included a 100-event \"train_sample\" if you want to explore the data before deciding to download the entire set.\n\n**Note:** Due to the number and size of the data files, we've only included the files from \"train_1\" in the Kernels environment (along with the full Test set, etc.)\n\nGood luck!", "321525": "Thank you to the organizers for setting up the competition. The data acquisition library is awesome and makes slicing a lot easier. From first glance having an understanding about the physics might give a little edge. Do you all think teams with more physicists will craft a better model, or will it come down to expertise in data science/ machine learning that cracks the tracks? Best luck to everyone!", "321268": "Hi organizers.\nThank you for arranging such a fascinating challenge!\nIt would be great if you could provide some graph or sketch of the detector which could help to understand a construction better. It would be helpful to relate certain features to the part of the detector. Thank you in advance.", "321293": "So this is the “big” dataset that we’ve been promised? I’m disappointed. It’s got nothing on TSA. :P", "321275": "Maybe these top 3 solutions from some inclass competition on tracks from OPERA experiment will be helpful for start.\nhttps://www.kaggle.com/c/dark-matter-signal-search-episode-1/discussion/36725", "322379": "This is very interesting challenge but I think we need to know physics alot....", "321271": "I missed out on the previous CERN competitions, glad I am not missing this one!", "374785": "The TrackML Kaggle competition has ended. Any participant (not necessarily with the highest score) who think they made valuable contributions with innovative algorithms (in particular if they have shared insights on the forum) are welcome to release publicly their code with an open source license and lightweight structured documentation. This will allow to compete for the « HEP meets ML » jury prizes: a NVidia V100 GPU, and two invitations to NIPS 2018 or the spring 2019 CERN workshop.See details in forum topic:\n \nhttps://www.kaggle.com/c/trackml-particle-identification/discussion/63289", "353645": "hi David I have a track_id attached to this message is ggplot of this. I was wondering is the tracker supposed to unify micro and macro events, because the only way I was able to make a positive tracker was using something like the Legendrian so the tracker is not north of strangeness and spontaneous. Does this come close to what you need? How do I complete my work if I do not have access to training sets 2-5? I called the plot something like tracking the destiny of future language because the \"macro\" events tether a Morse like line. I also think the curve looks like the two body Coulomb attraction/repulsion curve over distance. help me out if it is worth it and tell me what you think.", "322668": "Is that convinent with the very beginners?", "321485": "Looks interesting.", "321335": "" } }