![\"enter](\"https://raw.githubusercontent.com/takaiyuk/notebooks/master/output/score-zero-chart-181119.png\")
It has been a long time since LB became meaningless: we can see the LB filled with the score of 0.
\nYou will notice that there are many teams who submit only once scoring zero. So, I decided to scrape LB and to count the number.
The result is shown in the bar chart below.
\n\nThe number is 193 teams at the time I post this (November 19th 2018).
\nThis means that 193 teams will disappear or shake down due to their wrong submission when Private LB is calculated. (So, you may be now ranked 193 place higher than it is, if you are not lazy)
What do you think about this issue?
\n\n
I will update the chart the day before submission deadline.
", "messages": [ { "id": "424066", "postDate": "11/19/2018 13:53:26", "content": "It has been a long time since LB became meaningless: we can see the LB filled with the score of 0.
\nYou will notice that there are many teams who submit only once scoring zero. So, I decided to scrape LB and to count the number.
The result is shown in the bar chart below.
\n\nThe number is 193 teams at the time I post this (November 19th 2018).
\nThis means that 193 teams will disappear or shake down due to their wrong submission when Private LB is calculated. (So, you may be now ranked 193 place higher than it is, if you are not lazy)
What do you think about this issue?
\n\n
I will update the chart the day before submission deadline.
", "rawMarkdown": "It has been a long time since LB became meaningless: we can see the LB filled with the score of 0. \nYou will notice that there are many teams who submit only once scoring zero. So, I decided to scrape LB and to count the number. \n\n**The result is shown in the bar chart below.** \n\nThe number is **193 teams** at the time I post this (November 19th 2018). \nThis means that *193 teams will disappear or shake down due to their wrong submission when Private LB is calculated*. (So, you may be now ranked 193 place higher than it is, if you are not lazy) \n\nWhat do you think about this issue? \n\nThe fact, that those teams have only 1 submission doesn't necessarily mean they are not competing. I'm also competing in this competition, but I haven't do any submissions yet at all. Because all public test set target is available and there is no point to submit anything (only maybe to check submission format) - I'm simply validating my models offline. I'll probably submit only 2 submissions in the last day of the competition and those will be my final 2 submissions.
", "rawMarkdown": "The fact, that those teams have only 1 submission doesn't necessarily mean they are not competing. I'm also competing in this competition, but I haven't do any submissions yet at all. Because all public test set target is available and there is no point to submit anything (only maybe to check submission format) - I'm simply validating my models offline. I'll probably submit only 2 submissions in the last day of the competition and those will be my final 2 submissions.", "votes": null }, { "id": "424174", "postDate": "11/19/2018 17:43:53", "content": "Nice chart!
\n\nIt is too bad that it is now unclear which teams are performing well in the competition. I wonder why Kaggle won't clear the leaderboard. Maybe it is because the competition is nearly over.
", "rawMarkdown": "Nice chart!\n\nIt is too bad that it is now unclear which teams are performing well in the competition. I wonder why Kaggle won't clear the leaderboard. Maybe it is because the competition is nearly over.", "votes": null }, { "id": "424188", "postDate": "11/19/2018 18:03:57", "content": "I don't think it matters. Everyone knows not to rely on the LB in this competition.
", "rawMarkdown": "I don't think it matters. Everyone knows not to rely on the LB in this competition.", "votes": null }, { "id": "424333", "postDate": "11/19/2018 23:13:14", "content": "Thanks for your comments, @alijs, @Carlo and @Allen. Yes, as you all say, the number does not make sense at all. This chart is just for fun.\nAnd, as @alijs says, what we should do is just to focus on validating our models.
", "rawMarkdown": "Thanks for your comments, @alijs, @Carlo and @Allen. Yes, as you all say, the number does not make sense at all. This chart is just for fun.\nAnd, as @alijs says, what we should do is just to focus on validating our models.", "votes": null }, { "id": "424342", "postDate": "11/20/2018 00:04:24", "content": "I don't know if it's true but I tried to compute the rmse on test set and compare it to my score when I submit. The scores are real differents.
", "rawMarkdown": "I don't know if it's true but I tried to compute the rmse on test set and compare it to my score when I submit. The scores are real differents.", "votes": null }, { "id": "430525", "postDate": "11/30/2018 14:34:14", "content": "As I said, here are the chart (left) showing how many teams getting score 0 by the number of entries. There are 217 teams scoring 0 with one-shot.
\n\nAnd, I added another chart (right) showing how many teams getting score 2.1138 by the number of entries (You can get 2.1138 when all predict values are 0). There are 79 teams whose best score is 2.1138.
\n\nAnyway, once the competition ends, what you can do is to wait the result...like a New Year's Lottery!
\n\n![\"GACRP-LB-Chart-181130\"](\"https://raw.githubusercontent.com/takaiyuk/notebooks/master/output/GACRP-LB-Chart-181130.png\")
It would be really funny if the 1st place solution was the first person to submit an all-zeros submission.
", "rawMarkdown": "It would be really funny if the 1st place solution was the first person to submit an all-zeros submission.", "votes": null }, { "id": "430651", "postDate": "11/30/2018 19:41:50", "content": "The secret is to predict all zero's, except one.
", "rawMarkdown": "The secret is to predict all zero's, except one.", "votes": null }, { "id": "430682", "postDate": "11/30/2018 20:53:17", "content": "Yes, it is very funny, but it can happen.
", "rawMarkdown": "Yes, it is very funny, but it can happen.", "votes": null } ], "comments": [ { "id": 424168, "author_name": "alijs1", "author_url": "", "post_date": "11/19/2018 17:25:44", "content": "The fact, that those teams have only 1 submission doesn't necessarily mean they are not competing. I'm also competing in this competition, but I haven't do any submissions yet at all. Because all public test set target is available and there is no point to submit anything (only maybe to check submission format) - I'm simply validating my models offline. I'll probably submit only 2 submissions in the last day of the competition and those will be my final 2 submissions.
", "votes": null, "replies": [ { "id": 424342, "author_name": "qnkhuat", "author_url": "", "post_date": "11/20/2018 00:04:24", "content": "I don't know if it's true but I tried to compute the rmse on test set and compare it to my score when I submit. The scores are real differents.
", "votes": null, "replies": [] } ] }, { "id": 424174, "author_name": "carlolepelaars", "author_url": "", "post_date": "11/19/2018 17:43:53", "content": "Nice chart!
\n\nIt is too bad that it is now unclear which teams are performing well in the competition. I wonder why Kaggle won't clear the leaderboard. Maybe it is because the competition is nearly over.
", "votes": null, "replies": [] }, { "id": 424188, "author_name": "afgonczol", "author_url": "", "post_date": "11/19/2018 18:03:57", "content": "I don't think it matters. Everyone knows not to rely on the LB in this competition.
", "votes": null, "replies": [] }, { "id": 424333, "author_name": "takaishikawa", "author_url": "", "post_date": "11/19/2018 23:13:14", "content": "Thanks for your comments, @alijs, @Carlo and @Allen. Yes, as you all say, the number does not make sense at all. This chart is just for fun.\nAnd, as @alijs says, what we should do is just to focus on validating our models.
", "votes": null, "replies": [] }, { "id": 430525, "author_name": "takaishikawa", "author_url": "", "post_date": "11/30/2018 14:34:14", "content": "As I said, here are the chart (left) showing how many teams getting score 0 by the number of entries. There are 217 teams scoring 0 with one-shot.
\n\nAnd, I added another chart (right) showing how many teams getting score 2.1138 by the number of entries (You can get 2.1138 when all predict values are 0). There are 79 teams whose best score is 2.1138.
\n\nAnyway, once the competition ends, what you can do is to wait the result...like a New Year's Lottery!
\n\n
It would be really funny if the 1st place solution was the first person to submit an all-zeros submission.
", "votes": null, "replies": [] }, { "id": 430651, "author_name": "danisz", "author_url": "", "post_date": "11/30/2018 19:41:50", "content": "The secret is to predict all zero's, except one.
", "votes": null, "replies": [] }, { "id": 430682, "author_name": "takaishikawa", "author_url": "", "post_date": "11/30/2018 20:53:17", "content": "Yes, it is very funny, but it can happen.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "424066": "It has been a long time since LB became meaningless: we can see the LB filled with the score of 0. \nYou will notice that there are many teams who submit only once scoring zero. So, I decided to scrape LB and to count the number. \n\n**The result is shown in the bar chart below.** \n\nThe number is **193 teams** at the time I post this (November 19th 2018). \nThis means that *193 teams will disappear or shake down due to their wrong submission when Private LB is calculated*. (So, you may be now ranked 193 place higher than it is, if you are not lazy) \n\nWhat do you think about this issue? \n\n