I was not aware that the deadline for new entrants was 8th July. Also, i signed up for the competition approx. 20 days back only. So my results were not ready till then. Can you please extend the deadline for new entrants? And is there any way to extend the final submission date?
", "messages": [ { "id": "50182", "postDate": "07/09/2014 08:21:14", "content": "I was not aware that the deadline for new entrants was 8th July. Also, i signed up for the competition approx. 20 days back only. So my results were not ready till then. Can you please extend the deadline for new entrants? And is there any way to extend the final submission date?
", "rawMarkdown": "", "votes": null }, { "id": "50192", "postDate": "07/09/2014 12:03:31", "content": "Mainboard shows that competition will finish in 5,5 days, so I've tried to prepare first model to test data, but "competition is closed for new entrance".
\n\nI feel confused.
\n\nThanks for wasting my time ;)
", "rawMarkdown": "", "votes": null }, { "id": "50197", "postDate": "07/09/2014 13:20:22", "content": "Me too :(
\n\n\nI thought by accepting the terms & downloading the data, you were entered into the competition. At least, that's how it appears in the kaggle page flow.
\n\nWhen I went to submit my results, the competition was closed.
\n\nAny chance of extending the deadline ... again ?
", "rawMarkdown": "", "votes": null }, { "id": "50238", "postDate": "07/10/2014 04:59:00", "content": "To Kaggle admin,
\nEven if you are not able to extend the deadline, PLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition. Atleast, we have also spent time and effort on the problem and we atleast deserve this much.
\nWaiting in anticipation for the reply.
", "rawMarkdown": "", "votes": null }, { "id": "50256", "postDate": "07/10/2014 09:05:49", "content": "+1, forget about deadline and fail to submit. If those who provided benchmark submission or rather all-zeros submission when competition starts, can take part, but me not, I think such deadline is not good idea.
", "rawMarkdown": "", "votes": null }, { "id": "50257", "postDate": "07/10/2014 09:31:39", "content": "[quote=Trisco7;50238]
\nPLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition.
\n[/quote]
\nIf you wait just a few more days, you can do exactly this: submit and see your rank on private and public leaderboard, without scoring for the competition
\nIf you want to avoid this in the future: make an early submission. Even if it is random. Or just try out your models once in a while: working months on your first submission may very well lead to frustration.
\nFor those that do not understand the reasoning, or why other contestants may like this rule: It is really annoying if someone suddenly joins the competition in the last week and starts making top 10% submissions. It's not a nice feel, and may signal cheating. Also, with the number of contestants locked you'll know how much you should rank for top 10% or top 25%.
\nIt is probably too late now anyway to change the rules with 4.6 days to go. You risk annoying 999 contestants submitting within the rules to cater to a few latecomers. And if your model turns out to perform great, I am sure you can post the score and approach here, and it will garner the attention it deserves, even if it wasnt part of the competition.
\nIt's a bit of a bummer, but better luck next time!
", "rawMarkdown": "", "votes": null }, { "id": "50273", "postDate": "07/10/2014 14:40:41", "content": "" For those that do not understand the reasoning, or why other contestants may like this rule: It is really annoying if someone suddenly joins the competition in the last week and starts making top 10% submissions. It's not a nice feel, and may signal cheating. Also, with the number of contestants locked you'll know how much you should rank for top 10% or top 25%. "
\nEh, that logic is flawed, but I can see how it could work as a disincentive.
\nThere's nothing to prevent you making a random submission before the deadline, then saving your best model submission until the last week (or day).
\n> It's not a nice feel
\nThat's the feeling of competition :)
\n> may signal cheating.
\nHow so?
", "rawMarkdown": "", "votes": null }, { "id": "50275", "postDate": "07/10/2014 14:56:51", "content": "We have historically observed that the last few days of competitions are primetime for opportunistic cheaters (unsavory characters who create new accounts to either gain more submissions or buy more lottery tickets for the big draw). People who are too lazy to cross validate or plan their work ahead of time are generally also too lazy to plan their cheating ahead of time.
\nIt's not a perfect solution, but it is an effective deterrent against that itch to make just one more account when the clock is ticking down. As the old adage goes, "Locks are on doors to keep honest people honest."
", "rawMarkdown": "", "votes": null }, { "id": "50297", "postDate": "07/10/2014 17:41:34", "content": "I prefer the rule the way it is to the way it was. The last week used to be especially active for fake accounts. That was really bad last Fall. In Hashtag, I think I went from around 20th to the mid-40s...and then back to 23rd after cleanup, as the field size went from 250-300 to around 500 and back (I don't remember the exact numbers). Here's a thread about what was going on Hashtag submission spam. The effect of a surge like that is that, even if it does all get cleaned up, you just have no idea where you really are in the standings during the closing days.
Also, I think this way is better even without that issue. At some point in a competition, people should have to declare that they are in or out. A lot of the work in a Kaggle contest is done in the last week. At that point, you have to decide whether or not to put the time in to keep up. In making that decision, knowing who's in, who's out and what the field size is helps.
Just want to confirm that I can submit test results and see scores after the competition is finished? My model is not ready, and I am not really interested in participating the competition. But I want to see my score/rank on the full test data.
", "rawMarkdown": "", "votes": null }, { "id": "50316", "postDate": "07/10/2014 21:20:30", "content": "second to this:
\nPLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition.
", "rawMarkdown": "", "votes": null }, { "id": "50323", "postDate": "07/10/2014 23:25:17", "content": "Hi Kaggle Admin
I signed up around a month ago, and have been working on the problem; not wanting to submit random figures. I read the rules and all I could see with regards to submission was the following -
Submission Limits
\nThe end of the competition is the the 14th of July 2014 - I don't see how penalizing contestants for not submitting helps. As this is my first competition I was not aware of submit random values before x date rule to make sure you qualify at the end- I was also not sent an email notifying me of the importance of submitting before the 8th of July?
\nSeems I am in the same boat as a number of others - did we all just miss the requirement of forced submissions before the actual close of the competition?
\nThere is an argument of cheating and successful random submissions falling into the top positions, but surely these models would require validation via the client?
Regards
[quote=Kyle_BA;50323]
\n... I read the rules and all I could see with regards to submission was the following -
\nSubmission Limits
\n[/quote]
\nThe entry deadline is listed under Timeline.
", "rawMarkdown": "", "votes": null }, { "id": "50325", "postDate": "07/11/2014 00:12:35", "content": "Hi David
\nI remember seeing the 7/7 deadline for new entrants thinking I was already an entrant having signed up to the competition a month ago. Was the first submission deadline statement also there prior to the 7th?
\n7/7/2014 - Deadline for new entrants and team mergers
UPDATE: Extended to 7/8/2014 11:59 PM UTC - First Submission Deadline. Your team must make its first submission by this deadline. Team mergers are not allowed after this time.
Regards
", "rawMarkdown": "", "votes": null }, { "id": "50327", "postDate": "07/11/2014 00:57:52", "content": "[quote=Kyle_BA;50323]
\nThere is an argument of cheating and successful random submissions falling into the top positions, but surely these models would require validation via the client?
\nRegards
\n[/quote]
\nIts a very awkward question to answer transparently. Season competitors will answer you in a very terse manner. Why? Because we don't want to teach you to cheat
\nCheaters are common place in Kaggle competitions, many of them have made it to the top 10 and in contention to win prizes. The 1 week deadline is to restrict cheaters, who are contenders to win. Its a serious issue because they are really hard to detect. A few "less skilled" cheaters can be found on the leaderboard. Can you see them?
\nWith 3 more days to go, I only have 15 submissions left. I certainly would hope I have another extra 5 submissions. But like William says, gotta be honest, since everyone else is honest too.
", "rawMarkdown": "", "votes": null }, { "id": "50336", "postDate": "07/11/2014 07:45:12", "content": "how utterly frustrating this is. i, like several others, signed up and thought that that was all we needed to do with respect to having the opportunity to enter the competition prior to the stated 7/14 deadline. to now find that there was an interim deadline that wasn't stated in the DESCRIPTION nor RULES is extremely upsetting.
", "rawMarkdown": "", "votes": null }, { "id": "50354", "postDate": "07/11/2014 15:00:58", "content": "@Eu Jin Lok
\n"Cheaters are common place in Kaggle competitions, many of them have made it to the top 10 and in contention to win prizes. The 1 week deadline is to restrict cheaters, who are contenders to win. Its a serious issue because they are really hard to detect."
\nIsn't that what machine learning is for ?
", "rawMarkdown": "", "votes": null }, { "id": "50356", "postDate": "07/11/2014 15:17:43", "content": "To Kaggle_Admin:
\nWhy not to have a rule that: user can't accept terms and download data within one week before deadline? The frustration to many people as well as me is, there is no clear statement that I have to make a submission one week before actually deadline. I accepted the terms, I thought it means I am in the competition.
\nAnd existing rule can't avoid cheaters at all. A cheater can easily create 10 accounts, make a random submission for the 10 accounts at early stage. Isn't the right way to detect cheat by analyzing IP address and MAC address of the submission?
\nPlus, it seems I can't submit anymore after competition deadline. Why not? I assume most people here are for knowledge and experience. So if people can always make submission, more people are interested in this website. Even school professors may use some old competition as homework assignment.
\n\nThanks.
", "rawMarkdown": "", "votes": null }, { "id": "50365", "postDate": "07/11/2014 18:31:02", "content": "To Data Guru.
\nEven after the competition has ended you still have to accept the rules to download the data. For things like rules about redistribution and the like. So accepting the terms will always stay.
\n\nThe competition timeline speaks about a submission deadline, not an acept-the-rules deadline. Please carefully read the rules.
\n\nThis rule absolutely helps with catching and deterring cheaters. Kaggle would not add a rule like this if it was not effective or did not help them keep a birds-eye-view of the competition. Can you imagine the unexpected administrative load that would happen if 500 new contestants join a few hhours before the competition ends?
\n\nYou can absotely submit and receive a score after the competition ends! Just accept the rules and submit.
", "rawMarkdown": "", "votes": null }, { "id": "50371", "postDate": "07/11/2014 23:42:41", "content": "@kmcd
\nTouche!
", "rawMarkdown": "", "votes": null }, { "id": "76024", "postDate": "05/02/2015 03:42:55", "content": "Bottom Line. Dont penalize contestants that are here for the Knowledge...
\nLet people score their models and see how it stacks up against other approaches. Some are here to learn. There is no reason you cant simultaneously reward learners and penalize cheaters. Not sure why the rules dont allow for this flexibility.
", "rawMarkdown": "", "votes": null }, { "id": "76144", "postDate": "05/02/2015 20:13:29", "content": "Closed competitions are almost always open for submissions.
", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 50192, "author_name": "podludek", "author_url": "", "post_date": "07/09/2014 12:03:31", "content": "Mainboard shows that competition will finish in 5,5 days, so I've tried to prepare first model to test data, but "competition is closed for new entrance".
\n\nI feel confused.
\n\nThanks for wasting my time ;)
", "votes": null, "replies": [] }, { "id": 50197, "author_name": "kmcd191533", "author_url": "", "post_date": "07/09/2014 13:20:22", "content": "Me too :(
\n\n\nI thought by accepting the terms & downloading the data, you were entered into the competition. At least, that's how it appears in the kaggle page flow.
\n\nWhen I went to submit my results, the competition was closed.
\n\nAny chance of extending the deadline ... again ?
", "votes": null, "replies": [] }, { "id": 50238, "author_name": "trisco7", "author_url": "", "post_date": "07/10/2014 04:59:00", "content": "To Kaggle admin,
\nEven if you are not able to extend the deadline, PLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition. Atleast, we have also spent time and effort on the problem and we atleast deserve this much.
\nWaiting in anticipation for the reply.
", "votes": null, "replies": [] }, { "id": 50256, "author_name": "comanda", "author_url": "", "post_date": "07/10/2014 09:05:49", "content": "+1, forget about deadline and fail to submit. If those who provided benchmark submission or rather all-zeros submission when competition starts, can take part, but me not, I think such deadline is not good idea.
", "votes": null, "replies": [] }, { "id": 50257, "author_name": "triskelion", "author_url": "", "post_date": "07/10/2014 09:31:39", "content": "[quote=Trisco7;50238]
\nPLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition.
\n[/quote]
\nIf you wait just a few more days, you can do exactly this: submit and see your rank on private and public leaderboard, without scoring for the competition
\nIf you want to avoid this in the future: make an early submission. Even if it is random. Or just try out your models once in a while: working months on your first submission may very well lead to frustration.
\nFor those that do not understand the reasoning, or why other contestants may like this rule: It is really annoying if someone suddenly joins the competition in the last week and starts making top 10% submissions. It's not a nice feel, and may signal cheating. Also, with the number of contestants locked you'll know how much you should rank for top 10% or top 25%.
\nIt is probably too late now anyway to change the rules with 4.6 days to go. You risk annoying 999 contestants submitting within the rules to cater to a few latecomers. And if your model turns out to perform great, I am sure you can post the score and approach here, and it will garner the attention it deserves, even if it wasnt part of the competition.
\nIt's a bit of a bummer, but better luck next time!
", "votes": null, "replies": [] }, { "id": 50273, "author_name": "kmcd191533", "author_url": "", "post_date": "07/10/2014 14:40:41", "content": "" For those that do not understand the reasoning, or why other contestants may like this rule: It is really annoying if someone suddenly joins the competition in the last week and starts making top 10% submissions. It's not a nice feel, and may signal cheating. Also, with the number of contestants locked you'll know how much you should rank for top 10% or top 25%. "
\nEh, that logic is flawed, but I can see how it could work as a disincentive.
\nThere's nothing to prevent you making a random submission before the deadline, then saving your best model submission until the last week (or day).
\n> It's not a nice feel
\nThat's the feeling of competition :)
\n> may signal cheating.
\nHow so?
", "votes": null, "replies": [] }, { "id": 50275, "author_name": "wcukierski", "author_url": "", "post_date": "07/10/2014 14:56:51", "content": "We have historically observed that the last few days of competitions are primetime for opportunistic cheaters (unsavory characters who create new accounts to either gain more submissions or buy more lottery tickets for the big draw). People who are too lazy to cross validate or plan their work ahead of time are generally also too lazy to plan their cheating ahead of time.
\nIt's not a perfect solution, but it is an effective deterrent against that itch to make just one more account when the clock is ticking down. As the old adage goes, "Locks are on doors to keep honest people honest."
", "votes": null, "replies": [] }, { "id": 50297, "author_name": "davidthaler", "author_url": "", "post_date": "07/10/2014 17:41:34", "content": "I prefer the rule the way it is to the way it was. The last week used to be especially active for fake accounts. That was really bad last Fall. In Hashtag, I think I went from around 20th to the mid-40s...and then back to 23rd after cleanup, as the field size went from 250-300 to around 500 and back (I don't remember the exact numbers). Here's a thread about what was going on Hashtag submission spam. The effect of a surge like that is that, even if it does all get cleaned up, you just have no idea where you really are in the standings during the closing days.
Also, I think this way is better even without that issue. At some point in a competition, people should have to declare that they are in or out. A lot of the work in a Kaggle contest is done in the last week. At that point, you have to decide whether or not to put the time in to keep up. In making that decision, knowing who's in, who's out and what the field size is helps.
Just want to confirm that I can submit test results and see scores after the competition is finished? My model is not ready, and I am not really interested in participating the competition. But I want to see my score/rank on the full test data.
", "votes": null, "replies": [] }, { "id": 50316, "author_name": "dataguru0", "author_url": "", "post_date": "07/10/2014 21:20:30", "content": "second to this:
\nPLEASE allow the new entrants to submit their results and verify their respective score without considering them part of the competition.
", "votes": null, "replies": [] }, { "id": 50323, "author_name": "kyletaylor", "author_url": "", "post_date": "07/10/2014 23:25:17", "content": "Hi Kaggle Admin
I signed up around a month ago, and have been working on the problem; not wanting to submit random figures. I read the rules and all I could see with regards to submission was the following -
Submission Limits
\nThe end of the competition is the the 14th of July 2014 - I don't see how penalizing contestants for not submitting helps. As this is my first competition I was not aware of submit random values before x date rule to make sure you qualify at the end- I was also not sent an email notifying me of the importance of submitting before the 8th of July?
\nSeems I am in the same boat as a number of others - did we all just miss the requirement of forced submissions before the actual close of the competition?
\nThere is an argument of cheating and successful random submissions falling into the top positions, but surely these models would require validation via the client?
Regards
[quote=Kyle_BA;50323]
\n... I read the rules and all I could see with regards to submission was the following -
\nSubmission Limits
\n[/quote]
\nThe entry deadline is listed under Timeline.
", "votes": null, "replies": [] }, { "id": 50325, "author_name": "kyletaylor", "author_url": "", "post_date": "07/11/2014 00:12:35", "content": "Hi David
\nI remember seeing the 7/7 deadline for new entrants thinking I was already an entrant having signed up to the competition a month ago. Was the first submission deadline statement also there prior to the 7th?
\n7/7/2014 - Deadline for new entrants and team mergers
UPDATE: Extended to 7/8/2014 11:59 PM UTC - First Submission Deadline. Your team must make its first submission by this deadline. Team mergers are not allowed after this time.
Regards
", "votes": null, "replies": [] }, { "id": 50327, "author_name": "ejlok1", "author_url": "", "post_date": "07/11/2014 00:57:52", "content": "[quote=Kyle_BA;50323]
\nThere is an argument of cheating and successful random submissions falling into the top positions, but surely these models would require validation via the client?
\nRegards
\n[/quote]
\nIts a very awkward question to answer transparently. Season competitors will answer you in a very terse manner. Why? Because we don't want to teach you to cheat
\nCheaters are common place in Kaggle competitions, many of them have made it to the top 10 and in contention to win prizes. The 1 week deadline is to restrict cheaters, who are contenders to win. Its a serious issue because they are really hard to detect. A few "less skilled" cheaters can be found on the leaderboard. Can you see them?
\nWith 3 more days to go, I only have 15 submissions left. I certainly would hope I have another extra 5 submissions. But like William says, gotta be honest, since everyone else is honest too.
", "votes": null, "replies": [] }, { "id": 50336, "author_name": "cube43", "author_url": "", "post_date": "07/11/2014 07:45:12", "content": "how utterly frustrating this is. i, like several others, signed up and thought that that was all we needed to do with respect to having the opportunity to enter the competition prior to the stated 7/14 deadline. to now find that there was an interim deadline that wasn't stated in the DESCRIPTION nor RULES is extremely upsetting.
", "votes": null, "replies": [] }, { "id": 50354, "author_name": "kmcd191533", "author_url": "", "post_date": "07/11/2014 15:00:58", "content": "@Eu Jin Lok
\n"Cheaters are common place in Kaggle competitions, many of them have made it to the top 10 and in contention to win prizes. The 1 week deadline is to restrict cheaters, who are contenders to win. Its a serious issue because they are really hard to detect."
\nIsn't that what machine learning is for ?
", "votes": null, "replies": [] }, { "id": 50356, "author_name": "dataguru0", "author_url": "", "post_date": "07/11/2014 15:17:43", "content": "To Kaggle_Admin:
\nWhy not to have a rule that: user can't accept terms and download data within one week before deadline? The frustration to many people as well as me is, there is no clear statement that I have to make a submission one week before actually deadline. I accepted the terms, I thought it means I am in the competition.
\nAnd existing rule can't avoid cheaters at all. A cheater can easily create 10 accounts, make a random submission for the 10 accounts at early stage. Isn't the right way to detect cheat by analyzing IP address and MAC address of the submission?
\nPlus, it seems I can't submit anymore after competition deadline. Why not? I assume most people here are for knowledge and experience. So if people can always make submission, more people are interested in this website. Even school professors may use some old competition as homework assignment.
\n\nThanks.
", "votes": null, "replies": [] }, { "id": 50365, "author_name": "triskelion", "author_url": "", "post_date": "07/11/2014 18:31:02", "content": "To Data Guru.
\nEven after the competition has ended you still have to accept the rules to download the data. For things like rules about redistribution and the like. So accepting the terms will always stay.
\n\nThe competition timeline speaks about a submission deadline, not an acept-the-rules deadline. Please carefully read the rules.
\n\nThis rule absolutely helps with catching and deterring cheaters. Kaggle would not add a rule like this if it was not effective or did not help them keep a birds-eye-view of the competition. Can you imagine the unexpected administrative load that would happen if 500 new contestants join a few hhours before the competition ends?
\n\nYou can absotely submit and receive a score after the competition ends! Just accept the rules and submit.
", "votes": null, "replies": [] }, { "id": 50371, "author_name": "ejlok1", "author_url": "", "post_date": "07/11/2014 23:42:41", "content": "@kmcd
\nTouche!
", "votes": null, "replies": [] }, { "id": 76024, "author_name": "pemman", "author_url": "", "post_date": "05/02/2015 03:42:55", "content": "Bottom Line. Dont penalize contestants that are here for the Knowledge...
\nLet people score their models and see how it stacks up against other approaches. Some are here to learn. There is no reason you cant simultaneously reward learners and penalize cheaters. Not sure why the rules dont allow for this flexibility.
", "votes": null, "replies": [] }, { "id": 76144, "author_name": "wcukierski", "author_url": "", "post_date": "05/02/2015 20:13:29", "content": "Closed competitions are almost always open for submissions.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "50182": "", "50192": "", "50197": "", "50238": "", "50256": "", "50257": "", "50273": "", "50275": "", "50297": "", "50315": "", "50316": "", "50323": "", "50324": "", "50325": "", "50327": "", "50336": "", "50354": "", "50356": "", "50365": "", "50371": "", "76024": "", "76144": "" }, "source": "meta" }