Please stop making decisions with a high score! What time does this problem arise? Start kernels are enough. Someone is working in sweat, and with a few solutions you can get a higher position without much effort. This is not fair! Kernels are important for learning, so publish them after the competition! This is a competition, so respect other competitors as well
", "messages": [ { "id": "571295", "postDate": "07/09/2019 13:16:24", "content": "Please stop making decisions with a high score! What time does this problem arise? Start kernels are enough. Someone is working in sweat, and with a few solutions you can get a higher position without much effort. This is not fair! Kernels are important for learning, so publish them after the competition! This is a competition, so respect other competitors as well
", "rawMarkdown": "Please stop making decisions with a high score! What time does this problem arise? Start kernels are enough. Someone is working in sweat, and with a few solutions you can get a higher position without much effort. This is not fair! Kernels are important for learning, so publish them after the competition! This is a competition, so respect other competitors as well", "votes": null }, { "id": "571299", "postDate": "07/09/2019 13:24:47", "content": "We have two months to go so your take is not valid IMHO.
\n\nIf you are bothered about high scoring kernels beating you why don't you just blend them with your best models?
\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!
", "rawMarkdown": "We have two months to go so your take is not valid IMHO.\n\nIf you are bothered about high scoring kernels beating you why don't you just blend them with your best models?\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!", "votes": null }, { "id": "571309", "postDate": "07/09/2019 13:37:02", "content": "Your portion of medals you've already grabbed, you can afford to work and for interest. I'm trying to get medals and explore. There may be 2 months left, but soon it is not growing actively either. Do you want to have ~100 unique solutions and a lot of useless copies at the end? It will not help science, but the motivation on the carbide is discouraged. This problem swings from competition to competition
", "rawMarkdown": "Your portion of medals you've already grabbed, you can afford to work and for interest. I'm trying to get medals and explore. There may be 2 months left, but soon it is not growing actively either. Do you want to have ~100 unique solutions and a lot of useless copies at the end? It will not help science, but the motivation on the carbide is discouraged. This problem swings from competition to competition", "votes": null }, { "id": "571324", "postDate": "07/09/2019 13:53:15", "content": "You don't have to worry about improvement of science. Top scorers (~<-2.0) would not expose all of their strategies or features in the public kernel. What they (+ all other participants) publish is what they think is enough good to be shared among other competitors. I believe this makes some positive feedback for overall competition levels.
", "rawMarkdown": "You don't have to worry about improvement of science. Top scorers (~<-2.0) would not expose all of their strategies or features in the public kernel. What they (+ all other participants) publish is what they think is enough good to be shared among other competitors. I believe this makes some positive feedback for overall competition levels.", "votes": null }, { "id": "571329", "postDate": "07/09/2019 13:57:23", "content": "If you're trying to get medals but feel threatened by public kernels, you're going to have a bad time...
\n\nIf there is a public kernel which beats your score, take their ideas and use them to make improvements to your model.
", "rawMarkdown": "If you're trying to get medals but feel threatened by public kernels, you're going to have a bad time...\n\nIf there is a public kernel which beats your score, take their ideas and use them to make improvements to your model.", "votes": null }, { "id": "571367", "postDate": "07/09/2019 14:31:46", "content": "You don't need to worry about those high-scoring public kernel. Just focus on how you can improve your models. Take the idea coming from those kernels and integrate it. After all, there's a major shake-up in this competition. It's a battle of very good feature, model and cross-validation engineering.
", "rawMarkdown": "You don't need to worry about those high-scoring public kernel. Just focus on how you can improve your models. Take the idea coming from those kernels and integrate it. After all, there's a major shake-up in this competition. It's a battle of very good feature, model and cross-validation engineering.", "votes": null }, { "id": "571380", "postDate": "07/09/2019 14:45:07", "content": "From this topic I can conclude that author's main goal is to grab medals and use them for self-branding or other monkey business of same kind.\n *Hater mode active here *\nIf public kernels (best of which are not scoring high on the lb at the moment) beat your model, it just means than your model is khhhm-khm... bad. Stop thinking about any kind of self-branding while your decisions suck.\n *Hater mode off *\nTho I'd agree with this if it'd be the last week of competiton.\n> you can get a higher position without much effort. This is not fair!
", "rawMarkdown": "From this topic I can conclude that author's main goal is to grab medals and use them for self-branding or other monkey business of same kind.\n **Hater mode active here **\nIf public kernels (best of which are not scoring high on the lb at the moment) beat your model, it just means than your model is khhhm-khm... bad. Stop thinking about any kind of self-branding while your decisions suck.\n **Hater mode off **\nTho I'd agree with this if it'd be the last week of competiton.\n> you can get a higher position without much effort. This is not fair!", "votes": null }, { "id": "571400", "postDate": "07/09/2019 14:59:48", "content": "\n\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!
\n
And given the goal of the competition (and the papers to write ) it would be great that winning solutions were elegant/original yet simple solution rather than endless ensembling that won’t likely be very useful for the organizers.
", "rawMarkdown": "> I don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!\n\nAnd given the goal of the competition (and the papers to write ) it would be great that winning solutions were elegant/original yet simple solution rather than endless ensembling that won’t likely be very useful for the organizers.", "votes": null }, { "id": "571408", "postDate": "07/09/2019 15:09:56", "content": "It's not about score, it's about the competition. I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. It's not an isolated case when you can get into a medal by making a blend of public kernels. Do not forget this competition! It is worth respecting the efforts of others. Agree that if there was no problem, this topic would not be raised from competition to competition
", "rawMarkdown": "It's not about score, it's about the competition. I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. It's not an isolated case when you can get into a medal by making a blend of public kernels. Do not forget this competition! It is worth respecting the efforts of others. Agree that if there was no problem, this topic would not be raised from competition to competition", "votes": null }, { "id": "571427", "postDate": "07/09/2019 15:26:00", "content": "I spent a month too. So what? Gonna spend another two.\nPeople grab medals by making a blend of public kernels. So what? They're just more -l-a-z-y- cunning than those who didn't blend.\nNo one forbids you to blend too. It's your choise - to compete on your own or blend public solutions.\n> Do not forget this competition!
\n\nYes, but the main goal of this competition is not to win, but to solve important social problem.
", "rawMarkdown": "I spent a month too. So what? Gonna spend another two.\nPeople grab medals by making a blend of public kernels. So what? They're just more -l-a-z-y- cunning than those who didn't blend.\nNo one forbids you to blend too. It's your choise - to compete on your own or blend public solutions.\n> Do not forget this competition!\n\nYes, but the main goal of this competition is not to win, but to solve important social problem.", "votes": null }, { "id": "571433", "postDate": "07/09/2019 15:30:24", "content": "This is not a win-or-go-home game, it's more like you team up with other kagglers and make a bet. 'Let's work together and see who's the best'. \nYou compete with each others, yes that's true, but the most attractive part is you share same goal with many other genius kagglers. You can share their thoughts, you can figure out your position, you think and work, then you can be a better shit. \nKaggle is more like a sport rather than a race for me. Medal always matters, but dont let it take away the real joy.
", "rawMarkdown": "This is not a win-or-go-home game, it's more like you team up with other kagglers and make a bet. 'Let's work together and see who's the best'. \nYou compete with each others, yes that's true, but the most attractive part is you share same goal with many other genius kagglers. You can share their thoughts, you can figure out your position, you think and work, then you can be a better shit. \nKaggle is more like a sport rather than a race for me. Medal always matters, but dont let it take away the real joy.", "votes": null }, { "id": "571467", "postDate": "07/09/2019 16:10:58", "content": "\n\n\nSomeone is working in sweat
\n
.
\n\n\n\n\nand with a few solutions you can get a higher position without much effort
\n
Well, that means that someone does not work hard enough
", "rawMarkdown": "> Someone is working in sweat\n\n.\n> and with a few solutions you can get a higher position without much effort\n\n\nWell, that means that someone does not work hard enough", "votes": null }, { "id": "571571", "postDate": "07/09/2019 18:52:19", "content": "> I spent a month on my own, and now the question arises - why?
\n\nI think after one month you now understand the problem much better than if you start now. And after high score kernel you can learn what you have missed.
", "rawMarkdown": "> I spent a month on my own, and now the question arises - why? \n\nI think after one month you now understand the problem much better than if you start now. And after high score kernel you can learn what you have missed.", "votes": null }, { "id": "571942", "postDate": "07/10/2019 08:57:33", "content": "\n\n\nI spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend.
\n
You can always beat a public blend by adding your model to it ;)
", "rawMarkdown": "> I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. \n\nYou can always beat a public blend by adding your model to it ;)", "votes": null }, { "id": "573066", "postDate": "07/11/2019 18:36:07", "content": "Oh man! I think you're not understood the meaning of learning in kaggle. I already learned so much in this competition that even if I won the first place I would return the prize! Because the value acquired through learning was incalculable. I take this opportunity to thank the countless works published in this competition, especially @scirpus , @titericz, @artgor, @todnewman for you works.
", "rawMarkdown": "Oh man! I think you're not understood the meaning of learning in kaggle. I already learned so much in this competition that even if I won the first place I would return the prize! Because the value acquired through learning was incalculable. I take this opportunity to thank the countless works published in this competition, especially @scirpus , @titericz, @artgor, @todnewman for you works.", "votes": null }, { "id": "573085", "postDate": "07/11/2019 18:58:50", "content": "I commensurate with your topic idea choice. Been There. Done that.
", "rawMarkdown": "I commensurate with your topic idea choice. Been There. Done that.", "votes": null } ], "comments": [ { "id": 571299, "author_name": "scirpus", "author_url": "", "post_date": "07/09/2019 13:24:47", "content": "We have two months to go so your take is not valid IMHO.
\n\nIf you are bothered about high scoring kernels beating you why don't you just blend them with your best models?
\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!
", "votes": null, "replies": [ { "id": 571309, "author_name": "langsys", "author_url": "", "post_date": "07/09/2019 13:37:02", "content": "Your portion of medals you've already grabbed, you can afford to work and for interest. I'm trying to get medals and explore. There may be 2 months left, but soon it is not growing actively either. Do you want to have ~100 unique solutions and a lot of useless copies at the end? It will not help science, but the motivation on the carbide is discouraged. This problem swings from competition to competition
", "votes": null, "replies": [] }, { "id": 571329, "author_name": "joshxsarah", "author_url": "", "post_date": "07/09/2019 13:57:23", "content": "If you're trying to get medals but feel threatened by public kernels, you're going to have a bad time...
\n\nIf there is a public kernel which beats your score, take their ideas and use them to make improvements to your model.
", "votes": null, "replies": [] }, { "id": 571400, "author_name": "serigne", "author_url": "", "post_date": "07/09/2019 14:59:48", "content": "\n\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!
\n
And given the goal of the competition (and the papers to write ) it would be great that winning solutions were elegant/original yet simple solution rather than endless ensembling that won’t likely be very useful for the organizers.
", "votes": null, "replies": [] } ] }, { "id": 571324, "author_name": "leptony", "author_url": "", "post_date": "07/09/2019 13:53:15", "content": "You don't have to worry about improvement of science. Top scorers (~<-2.0) would not expose all of their strategies or features in the public kernel. What they (+ all other participants) publish is what they think is enough good to be shared among other competitors. I believe this makes some positive feedback for overall competition levels.
", "votes": null, "replies": [] }, { "id": 571367, "author_name": "projdev", "author_url": "", "post_date": "07/09/2019 14:31:46", "content": "You don't need to worry about those high-scoring public kernel. Just focus on how you can improve your models. Take the idea coming from those kernels and integrate it. After all, there's a major shake-up in this competition. It's a battle of very good feature, model and cross-validation engineering.
", "votes": null, "replies": [] }, { "id": 571380, "author_name": "stanislavblinov", "author_url": "", "post_date": "07/09/2019 14:45:07", "content": "From this topic I can conclude that author's main goal is to grab medals and use them for self-branding or other monkey business of same kind.\n *Hater mode active here *\nIf public kernels (best of which are not scoring high on the lb at the moment) beat your model, it just means than your model is khhhm-khm... bad. Stop thinking about any kind of self-branding while your decisions suck.\n *Hater mode off *\nTho I'd agree with this if it'd be the last week of competiton.\n> you can get a higher position without much effort. This is not fair!
", "votes": null, "replies": [ { "id": 571408, "author_name": "langsys", "author_url": "", "post_date": "07/09/2019 15:09:56", "content": "It's not about score, it's about the competition. I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. It's not an isolated case when you can get into a medal by making a blend of public kernels. Do not forget this competition! It is worth respecting the efforts of others. Agree that if there was no problem, this topic would not be raised from competition to competition
", "votes": null, "replies": [] }, { "id": 571427, "author_name": "stanislavblinov", "author_url": "", "post_date": "07/09/2019 15:26:00", "content": "I spent a month too. So what? Gonna spend another two.\nPeople grab medals by making a blend of public kernels. So what? They're just more -l-a-z-y- cunning than those who didn't blend.\nNo one forbids you to blend too. It's your choise - to compete on your own or blend public solutions.\n> Do not forget this competition!
\n\nYes, but the main goal of this competition is not to win, but to solve important social problem.
", "votes": null, "replies": [] }, { "id": 571571, "author_name": "sergeyzlobin", "author_url": "", "post_date": "07/09/2019 18:52:19", "content": "> I spent a month on my own, and now the question arises - why?
\n\nI think after one month you now understand the problem much better than if you start now. And after high score kernel you can learn what you have missed.
", "votes": null, "replies": [] }, { "id": 571942, "author_name": "cpmpml", "author_url": "", "post_date": "07/10/2019 08:57:33", "content": "\n\n\nI spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend.
\n
You can always beat a public blend by adding your model to it ;)
", "votes": null, "replies": [] } ] }, { "id": 571433, "author_name": "littlebeandog", "author_url": "", "post_date": "07/09/2019 15:30:24", "content": "This is not a win-or-go-home game, it's more like you team up with other kagglers and make a bet. 'Let's work together and see who's the best'. \nYou compete with each others, yes that's true, but the most attractive part is you share same goal with many other genius kagglers. You can share their thoughts, you can figure out your position, you think and work, then you can be a better shit. \nKaggle is more like a sport rather than a race for me. Medal always matters, but dont let it take away the real joy.
", "votes": null, "replies": [] }, { "id": 571467, "author_name": "", "author_url": "", "post_date": "07/09/2019 16:10:58", "content": "\n\n\nSomeone is working in sweat
\n
.
\n\n\n\n\nand with a few solutions you can get a higher position without much effort
\n
Well, that means that someone does not work hard enough
", "votes": null, "replies": [] }, { "id": 573066, "author_name": "adrianoavelar", "author_url": "", "post_date": "07/11/2019 18:36:07", "content": "Oh man! I think you're not understood the meaning of learning in kaggle. I already learned so much in this competition that even if I won the first place I would return the prize! Because the value acquired through learning was incalculable. I take this opportunity to thank the countless works published in this competition, especially @scirpus , @titericz, @artgor, @todnewman for you works.
", "votes": null, "replies": [] }, { "id": 573085, "author_name": "stocks", "author_url": "", "post_date": "07/11/2019 18:58:50", "content": "I commensurate with your topic idea choice. Been There. Done that.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "571295": "Please stop making decisions with a high score! What time does this problem arise? Start kernels are enough. Someone is working in sweat, and with a few solutions you can get a higher position without much effort. This is not fair! Kernels are important for learning, so publish them after the competition! This is a competition, so respect other competitors as well", "571299": "We have two months to go so your take is not valid IMHO.\n\nIf you are bothered about high scoring kernels beating you why don't you just blend them with your best models?\n\nI don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!", "571309": "Your portion of medals you've already grabbed, you can afford to work and for interest. I'm trying to get medals and explore. There may be 2 months left, but soon it is not growing actively either. Do you want to have ~100 unique solutions and a lot of useless copies at the end? It will not help science, but the motivation on the carbide is discouraged. This problem swings from competition to competition", "571324": "You don't have to worry about improvement of science. Top scorers (~<-2.0) would not expose all of their strategies or features in the public kernel. What they (+ all other participants) publish is what they think is enough good to be shared among other competitors. I believe this makes some positive feedback for overall competition levels.", "571329": "If you're trying to get medals but feel threatened by public kernels, you're going to have a bad time...\n\nIf there is a public kernel which beats your score, take their ideas and use them to make improvements to your model.", "571367": "You don't need to worry about those high-scoring public kernel. Just focus on how you can improve your models. Take the idea coming from those kernels and integrate it. After all, there's a major shake-up in this competition. It's a battle of very good feature, model and cross-validation engineering.", "571380": "From this topic I can conclude that author's main goal is to grab medals and use them for self-branding or other monkey business of same kind.\n **Hater mode active here **\nIf public kernels (best of which are not scoring high on the lb at the moment) beat your model, it just means than your model is khhhm-khm... bad. Stop thinking about any kind of self-branding while your decisions suck.\n **Hater mode off **\nTho I'd agree with this if it'd be the last week of competiton.\n> you can get a higher position without much effort. This is not fair!", "571400": "> I don't care about my position as the goal here is to enable scientists to improve their calculations using traditional machine learning techniques. Just take pleasure in helping them out and you will learn a lot from the journey!\n\nAnd given the goal of the competition (and the papers to write ) it would be great that winning solutions were elegant/original yet simple solution rather than endless ensembling that won’t likely be very useful for the organizers.", "571408": "It's not about score, it's about the competition. I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. It's not an isolated case when you can get into a medal by making a blend of public kernels. Do not forget this competition! It is worth respecting the efforts of others. Agree that if there was no problem, this topic would not be raised from competition to competition", "571427": "I spent a month too. So what? Gonna spend another two.\nPeople grab medals by making a blend of public kernels. So what? They're just more -l-a-z-y- cunning than those who didn't blend.\nNo one forbids you to blend too. It's your choise - to compete on your own or blend public solutions.\n> Do not forget this competition!\n\nYes, but the main goal of this competition is not to win, but to solve important social problem.", "571433": "This is not a win-or-go-home game, it's more like you team up with other kagglers and make a bet. 'Let's work together and see who's the best'. \nYou compete with each others, yes that's true, but the most attractive part is you share same goal with many other genius kagglers. You can share their thoughts, you can figure out your position, you think and work, then you can be a better shit. \nKaggle is more like a sport rather than a race for me. Medal always matters, but dont let it take away the real joy.", "571467": "> Someone is working in sweat\n\n.\n> and with a few solutions you can get a higher position without much effort\n\n\nWell, that means that someone does not work hard enough", "571571": "> I spent a month on my own, and now the question arises - why? \n\nI think after one month you now understand the problem much better than if you start now. And after high score kernel you can learn what you have missed.", "571942": "> I spent a month on my own, and now the question arises - why? If it was possible to wait, and then make a blend. \n\nYou can always beat a public blend by adding your model to it ;)", "573066": "Oh man! I think you're not understood the meaning of learning in kaggle. I already learned so much in this competition that even if I won the first place I would return the prize! Because the value acquired through learning was incalculable. I take this opportunity to thank the countless works published in this competition, especially @scirpus , @titericz, @artgor, @todnewman for you works.", "573085": "I commensurate with your topic idea choice. Been There. Done that." }, "source": "meta" }