The only way anyone would have time to do this if they were unemployed. And only experienced data scientists with a very high skill set are capable of doing this at all. And if you are an experienced data scientist with a very high skill set, you almost certainly have a job and will therefore have no time to do this by the deadline. So who exactly is Kaggle targeting with this? College students with no jobs who have tons of time on their hands? They won't have the know-how or experience to do this. Data scientists with worlds of experience who happen to not have a job right now? What's the Venn diagram of \"unemployed\" and \"experienced in high-demand data scientist\"? Pretty much zero, right? So I don't see how this is possible to get done by the deadline, unless you are literally a 20-years-experienced data scientist who is so rich from all his data science work that he can afford to not take a salary for 1-2 years and just spends his days doing Kaggle competitions. Is THAT the demographic Kaggle is targeting with this?
\n\nSeriously, not even trying to be rude here, but people have jobs. I'm personally very interested in this analysis and want to see what I can do with it. But I have a job that takes up 40+ hours of my week, I have deadlines to meet at my job, need to work into the evenings a lot, which means I would only have weekends for this at all, and I really don't want to spend my weekends staring at code and databases like I do all week. Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.
\n\nAND I might add, this competition is going on smack-dab in the middle of the holiday season. Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?
", "messages": [ { "id": "439511", "postDate": "12/15/2018 16:58:00", "content": "The only way anyone would have time to do this if they were unemployed. And only experienced data scientists with a very high skill set are capable of doing this at all. And if you are an experienced data scientist with a very high skill set, you almost certainly have a job and will therefore have no time to do this by the deadline. So who exactly is Kaggle targeting with this? College students with no jobs who have tons of time on their hands? They won't have the know-how or experience to do this. Data scientists with worlds of experience who happen to not have a job right now? What's the Venn diagram of \"unemployed\" and \"experienced in high-demand data scientist\"? Pretty much zero, right? So I don't see how this is possible to get done by the deadline, unless you are literally a 20-years-experienced data scientist who is so rich from all his data science work that he can afford to not take a salary for 1-2 years and just spends his days doing Kaggle competitions. Is THAT the demographic Kaggle is targeting with this?
\n\nSeriously, not even trying to be rude here, but people have jobs. I'm personally very interested in this analysis and want to see what I can do with it. But I have a job that takes up 40+ hours of my week, I have deadlines to meet at my job, need to work into the evenings a lot, which means I would only have weekends for this at all, and I really don't want to spend my weekends staring at code and databases like I do all week. Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.
\n\nAND I might add, this competition is going on smack-dab in the middle of the holiday season. Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?
", "rawMarkdown": "The only way anyone would have time to do this if they were unemployed. And only experienced data scientists with a very high skill set are capable of doing this at all. And if you are an experienced data scientist with a very high skill set, you almost certainly have a job and will therefore have no time to do this by the deadline. So who exactly is Kaggle targeting with this? College students with no jobs who have tons of time on their hands? They won't have the know-how or experience to do this. Data scientists with worlds of experience who happen to not have a job right now? What's the Venn diagram of \"unemployed\" and \"experienced in high-demand data scientist\"? Pretty much zero, right? So I don't see how this is possible to get done by the deadline, unless you are literally a 20-years-experienced data scientist who is so rich from all his data science work that he can afford to not take a salary for 1-2 years and just spends his days doing Kaggle competitions. Is THAT the demographic Kaggle is targeting with this?\n\nSeriously, not even trying to be rude here, but people have jobs. I'm personally very interested in this analysis and want to see what I can do with it. But I have a job that takes up 40+ hours of my week, I have deadlines to meet at my job, need to work into the evenings a lot, which means I would only have weekends for this at all, and I really don't want to spend my weekends staring at code and databases like I do all week. Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.\n\nAND I might add, this competition is going on smack-dab in the middle of the holiday season. Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?", "votes": null }, { "id": "439726", "postDate": "12/16/2018 08:02:26", "content": "Then be quiet and don't do it. Some of the greatest ideas in the world have come from unexpected people, and to potentially impact a real change in the NFL would be an honor to some.
", "rawMarkdown": "Then be quiet and don't do it. Some of the greatest ideas in the world have come from unexpected people, and to potentially impact a real change in the NFL would be an honor to some.", "votes": null }, { "id": "439986", "postDate": "12/16/2018 19:59:44", "content": "\n\n\nRegardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?
\n
Yes. We do it every year. Some of us need a Christmas miracle to get us into the top 10 and that miracle can only happen if we make submissions.
", "rawMarkdown": ">Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?\n\nYes. We do it every year. Some of us need a Christmas miracle to get us into the top 10 and that miracle can only happen if we make submissions.", "votes": null }, { "id": "440012", "postDate": "12/16/2018 21:33:35", "content": "You'd rather do this than spend time with your family? Then you need to get your priorities straight. Seriously. I'm not saying this isn't a cool thing to do, it is, but it is not more important than family. Period.
", "rawMarkdown": "You'd rather do this than spend time with your family? Then you need to get your priorities straight. Seriously. I'm not saying this isn't a cool thing to do, it is, but it is not more important than family. Period.", "votes": null }, { "id": "440049", "postDate": "12/17/2018 00:42:15", "content": "Nah, I spend time with them on Christmas and then do some Kaggling before bed.
", "rawMarkdown": "Nah, I spend time with them on Christmas and then do some Kaggling before bed.", "votes": null }, { "id": "440065", "postDate": "12/17/2018 01:34:16", "content": "I think you're overestimating the level of effort involved here from a Data Science stand point. The winner will likely be a n analyst with domain knowledge. For instance a pretty obvious solution to this problem would involve mandating a fair catch since runbacks are involved in most concussions, but that would never be an acceptable solution in the NFL since it's not going to maintain interesting games while also decreasing the frequency of concussion.
", "rawMarkdown": "I think you're overestimating the level of effort involved here from a Data Science stand point. The winner will likely be a n analyst with domain knowledge. For instance a pretty obvious solution to this problem would involve mandating a fair catch since runbacks are involved in most concussions, but that would never be an acceptable solution in the NFL since it's not going to maintain interesting games while also decreasing the frequency of concussion.", "votes": null }, { "id": "440097", "postDate": "12/17/2018 03:09:23", "content": "I definitely don't see the issue with spending my holidays on Kaggle. I think it's a great way to spend my free time to be honest!
", "rawMarkdown": "I definitely don't see the issue with spending my holidays on Kaggle. I think it's a great way to spend my free time to be honest!", "votes": null }, { "id": "441405", "postDate": "12/18/2018 16:14:28", "content": "\n\n\nPlus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.
\n
Prove it
", "rawMarkdown": "> Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.\n\nProve it", "votes": null }, { "id": "441870", "postDate": "12/19/2018 07:23:13", "content": "False. If you had a solid domain knowledge you'd understand that analyzing the vast majority of this data is not necessary to come up with a good suggestion.
\n\nThis is a loosely defined problem, and therefore inherently requires more creativity to solve. For a data scientist with a day job, it is nice to see a problem that doesn't only require a narrow domain knowledge and the hardware to iterate through tons of hyperaprameter combinations. If you're a statistician, you should know that fighting for a .978 AUC vs a .977 AUC is almost certainly fighting to get lucky on some blip of noise compared to the person with an .977 AUC.
", "rawMarkdown": "False. If you had a solid domain knowledge you'd understand that analyzing the vast majority of this data is not necessary to come up with a good suggestion. \n\nThis is a loosely defined problem, and therefore inherently requires more creativity to solve. For a data scientist with a day job, it is nice to see a problem that doesn't only require a narrow domain knowledge and the hardware to iterate through tons of hyperaprameter combinations. If you're a statistician, you should know that fighting for a .978 AUC vs a .977 AUC is almost certainly fighting to get lucky on some blip of noise compared to the person with an .977 AUC.", "votes": null }, { "id": "445288", "postDate": "12/26/2018 06:12:03", "content": "I think this type of problem is much more interesting than getting that last .001 accuracy. You have to really understand your data to make reasonable recommendations. It also involves getting some domain knowledge into the weeds that you never would have done otherwise. If you think it's too time consuming and unpleasant you should get another interest. You don't have to be a CS PhD to do this kind of analysis.
", "rawMarkdown": "I think this type of problem is much more interesting than getting that last .001 accuracy. You have to really understand your data to make reasonable recommendations. It also involves getting some domain knowledge into the weeds that you never would have done otherwise. If you think it's too time consuming and unpleasant you should get another interest. You don't have to be a CS PhD to do this kind of analysis.", "votes": null } ], "comments": [ { "id": 439726, "author_name": "gregalytics", "author_url": "", "post_date": "12/16/2018 08:02:26", "content": "Then be quiet and don't do it. Some of the greatest ideas in the world have come from unexpected people, and to potentially impact a real change in the NFL would be an honor to some.
", "votes": null, "replies": [] }, { "id": 439986, "author_name": "brandenkmurray", "author_url": "", "post_date": "12/16/2018 19:59:44", "content": "\n\n\nRegardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?
\n
Yes. We do it every year. Some of us need a Christmas miracle to get us into the top 10 and that miracle can only happen if we make submissions.
", "votes": null, "replies": [ { "id": 440012, "author_name": "arjunbhalla675", "author_url": "", "post_date": "12/16/2018 21:33:35", "content": "You'd rather do this than spend time with your family? Then you need to get your priorities straight. Seriously. I'm not saying this isn't a cool thing to do, it is, but it is not more important than family. Period.
", "votes": null, "replies": [] }, { "id": 440049, "author_name": "brandenkmurray", "author_url": "", "post_date": "12/17/2018 00:42:15", "content": "Nah, I spend time with them on Christmas and then do some Kaggling before bed.
", "votes": null, "replies": [] } ] }, { "id": 440065, "author_name": "naldershof", "author_url": "", "post_date": "12/17/2018 01:34:16", "content": "I think you're overestimating the level of effort involved here from a Data Science stand point. The winner will likely be a n analyst with domain knowledge. For instance a pretty obvious solution to this problem would involve mandating a fair catch since runbacks are involved in most concussions, but that would never be an acceptable solution in the NFL since it's not going to maintain interesting games while also decreasing the frequency of concussion.
", "votes": null, "replies": [] }, { "id": 440097, "author_name": "delayedkarma", "author_url": "", "post_date": "12/17/2018 03:09:23", "content": "I definitely don't see the issue with spending my holidays on Kaggle. I think it's a great way to spend my free time to be honest!
", "votes": null, "replies": [] }, { "id": 441405, "author_name": "serigne", "author_url": "", "post_date": "12/18/2018 16:14:28", "content": "\n\n\nPlus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.
\n
Prove it
", "votes": null, "replies": [] }, { "id": 441870, "author_name": "colin33", "author_url": "", "post_date": "12/19/2018 07:23:13", "content": "False. If you had a solid domain knowledge you'd understand that analyzing the vast majority of this data is not necessary to come up with a good suggestion.
\n\nThis is a loosely defined problem, and therefore inherently requires more creativity to solve. For a data scientist with a day job, it is nice to see a problem that doesn't only require a narrow domain knowledge and the hardware to iterate through tons of hyperaprameter combinations. If you're a statistician, you should know that fighting for a .978 AUC vs a .977 AUC is almost certainly fighting to get lucky on some blip of noise compared to the person with an .977 AUC.
", "votes": null, "replies": [] }, { "id": 445288, "author_name": "dave1216", "author_url": "", "post_date": "12/26/2018 06:12:03", "content": "I think this type of problem is much more interesting than getting that last .001 accuracy. You have to really understand your data to make reasonable recommendations. It also involves getting some domain knowledge into the weeds that you never would have done otherwise. If you think it's too time consuming and unpleasant you should get another interest. You don't have to be a CS PhD to do this kind of analysis.
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "439511": "The only way anyone would have time to do this if they were unemployed. And only experienced data scientists with a very high skill set are capable of doing this at all. And if you are an experienced data scientist with a very high skill set, you almost certainly have a job and will therefore have no time to do this by the deadline. So who exactly is Kaggle targeting with this? College students with no jobs who have tons of time on their hands? They won't have the know-how or experience to do this. Data scientists with worlds of experience who happen to not have a job right now? What's the Venn diagram of \"unemployed\" and \"experienced in high-demand data scientist\"? Pretty much zero, right? So I don't see how this is possible to get done by the deadline, unless you are literally a 20-years-experienced data scientist who is so rich from all his data science work that he can afford to not take a salary for 1-2 years and just spends his days doing Kaggle competitions. Is THAT the demographic Kaggle is targeting with this?\n\nSeriously, not even trying to be rude here, but people have jobs. I'm personally very interested in this analysis and want to see what I can do with it. But I have a job that takes up 40+ hours of my week, I have deadlines to meet at my job, need to work into the evenings a lot, which means I would only have weekends for this at all, and I really don't want to spend my weekends staring at code and databases like I do all week. Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.\n\nAND I might add, this competition is going on smack-dab in the middle of the holiday season. Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?", "439726": "Then be quiet and don't do it. Some of the greatest ideas in the world have come from unexpected people, and to potentially impact a real change in the NFL would be an honor to some.", "439986": ">Regardless of who you are, do you REALLY want to spend your Christmas-New Years period crunching numbers and hacking out data science code round the clock, instead of spending time with your family?\n\nYes. We do it every year. Some of us need a Christmas miracle to get us into the top 10 and that miracle can only happen if we make submissions.", "440012": "You'd rather do this than spend time with your family? Then you need to get your priorities straight. Seriously. I'm not saying this isn't a cool thing to do, it is, but it is not more important than family. Period.", "440049": "Nah, I spend time with them on Christmas and then do some Kaggling before bed.", "440065": "I think you're overestimating the level of effort involved here from a Data Science stand point. The winner will likely be a n analyst with domain knowledge. For instance a pretty obvious solution to this problem would involve mandating a fair catch since runbacks are involved in most concussions, but that would never be an acceptable solution in the NFL since it's not going to maintain interesting games while also decreasing the frequency of concussion.", "440097": "I definitely don't see the issue with spending my holidays on Kaggle. I think it's a great way to spend my free time to be honest!", "441405": "> Plus the winner of this competition will obviously be some 50-year-old data scientist with multiple PhDs and 20-25 years experience who works/worked at Amazon/Google/Microsoft/etc. and I or most people entering this are not that person.\n\nProve it", "441870": "False. If you had a solid domain knowledge you'd understand that analyzing the vast majority of this data is not necessary to come up with a good suggestion. \n\nThis is a loosely defined problem, and therefore inherently requires more creativity to solve. For a data scientist with a day job, it is nice to see a problem that doesn't only require a narrow domain knowledge and the hardware to iterate through tons of hyperaprameter combinations. If you're a statistician, you should know that fighting for a .978 AUC vs a .977 AUC is almost certainly fighting to get lucky on some blip of noise compared to the person with an .977 AUC.", "445288": "I think this type of problem is much more interesting than getting that last .001 accuracy. You have to really understand your data to make reasonable recommendations. It also involves getting some domain knowledge into the weeds that you never would have done otherwise. If you think it's too time consuming and unpleasant you should get another interest. You don't have to be a CS PhD to do this kind of analysis." }, "source": "meta" }