Thank you everyone for a great competition! I wanted to share a few observations given that this format was a \"first\". Of course, anything I share is merely an opinion that may or may not have helped in this case.
\n\nInformation vs Exploration. I used an informational approach for my main kernel since this was a report. To me that means:
\n\nTargeted Analysis. This was a great case of a lot of data and very little time. I found that a logic tree and some initial exploration helped focus where I wanted to work. It's generally better to prioritize on one thing and kill it vs. spreading your effort across too many opportunities.
\n\nFunctional Style. I spent a lot of time learning how to make my notebook look better. Not just for flair, but to help guide the reader. Section headings, colored fonts, anything that might enhance readability. R users have an advantage with R Markdown reports that are relatively easy to style. Regular notebooks require more manual HTML.
\n\nIt would great to hear any observations/challenges/etc that others might have!
", "messages": [ { "id": 458472, "postDate": "2019-01-19T18:22:48.280Z", "content": "Thank you everyone for a great competition! I wanted to share a few observations given that this format was a \"first\". Of course, anything I share is merely an opinion that may or may not have helped in this case.
\n\nInformation vs Exploration. I used an informational approach for my main kernel since this was a report. To me that means:
\n\nTargeted Analysis. This was a great case of a lot of data and very little time. I found that a logic tree and some initial exploration helped focus where I wanted to work. It's generally better to prioritize on one thing and kill it vs. spreading your effort across too many opportunities.
\n\nFunctional Style. I spent a lot of time learning how to make my notebook look better. Not just for flair, but to help guide the reader. Section headings, colored fonts, anything that might enhance readability. R users have an advantage with R Markdown reports that are relatively easy to style. Regular notebooks require more manual HTML.
\n\nIt would great to hear any observations/challenges/etc that others might have!
", "rawMarkdown": "Thank you everyone for a great competition! I wanted to share a few observations given that this format was a \"first\". Of course, anything I share is merely an opinion that may or may not have helped in this case.\n\n**Information vs Exploration.** I used an informational approach for my main kernel since this was a report. To me that means:\n\n - Bottom line/summary up front\n - Clear outline for the content\n - Conclusions and relevant evidence only\n - Plots with a specific message\n\n**Targeted Analysis.** This was a great case of a lot of data and very little time. I found that a logic tree and some initial exploration helped focus where I wanted to work. It's generally better to prioritize on one thing and kill it vs. spreading your effort across too many opportunities.\n\n**Functional Style.** I spent a lot of time learning how to make my notebook look better. Not just for flair, but to help guide the reader. Section headings, colored fonts, anything that might enhance readability. R users have an advantage with R Markdown reports that are relatively easy to style. Regular notebooks require more manual HTML.\n\nIt would great to hear any observations/challenges/etc that others might have!", "votes": 17 }, { "id": 460005, "postDate": "2019-01-22T19:50:28.373Z", "content": "The deleted comment had some great ideas about how to think about this competition. Framing it as presentation to senior level executives was obviously the right way to approach the problem. There were some other helpful tips in there, but I don't recall them.
\n\nRe-reading the submission criteria, I would tweak it for the next analytics competition.
\n\n\nA valid submission will include:\n\n\n
\nSummary Slides: A summary of the proposed rule change presented in slide format and delivered as either a PDF or PPT file.\n
\nKernels Analysis: At least one kernel containing the analysis to support your proposal. All kernels submitted must be made public on or before the submission deadline to be eligible. If submitting as a team, all team members must be listed as collaborators on all kernels submitted.\n
\nSubmissions will be judged by the NFL on the following criteria:\n
\nSolution Efficacy - Have you clearly demonstrated, through your data analysis, that you have an understanding of what play features may be associated with concussions and how your proposed rule change(s) will reduce these injuries? Your kernels should be easy to understand and the analysis should be reproducible.\n
\nGame Integrity - Is your proposal actionable by the NFL? Could they implement your rule change and still maintain the integrity of the game? Have you considered the way your proposed changes to game dynamics could introduce new risks to player safety? Strong submissions will demonstrate an understanding for the game overall.\n
I would rephrase the top part (based on my understanding of the judging)
\n\nPresentation: A 5 to 10 minute presentation of the proposed rule change presented in slide format and delivered as either a PDF or PPT file. The audience is senior non-technical management (VP's, directors, etc). It would be helpful but not required to have an appendix which references a kernel to every graph, calculation, and assumption.
\n\nKernels Analysis: Consider this supplementary material. The senior management will select a handful of presentations and have technical experts review the kernels to validate the calculations and assumptions. These kernel(s) should not be a record of your exploration but only the justification for the statements in the presentation. Kernels in your submission will automatically turn public at the contest deadline, but feel free to share them early and often. Instead of one big kernel, consider having multiple small kernels each tackling a specific topic. Make sure to use Kaggle's new feature that allows users to use output of one kernel as input to the next kernel.
\n\nFor the judging criteria, I would add something like
\n\nClarity/Concise -- Are the slides professional quality? Do they stand alone without narration? Are there any weak slides that should have been removed? Are the ideas clear and unambiguous?
\n\nAnyway, that's my 2 cents. I'm not intending to be a sour grape. I congratulate all the winners. They did a fantastic job. With a little guidance, I think we can collective raise the bar for the next competition.
", "rawMarkdown": "The deleted comment had some great ideas about how to think about this competition. Framing it as presentation to senior level executives was obviously the right way to approach the problem. There were some other helpful tips in there, but I don't recall them.\n\nRe-reading the submission criteria, I would tweak it for the next analytics competition.\n\n\nA valid submission will include:\n\n\n\nI would rephrase the top part (based on my understanding of the judging)\n\nPresentation: A 5 to 10 minute presentation of the proposed rule change presented in slide format and delivered as either a PDF or PPT file. The audience is senior non-technical management (VP's, directors, etc). It would be helpful but not required to have an appendix which references a kernel to every graph, calculation, and assumption.\n\nKernels Analysis: Consider this supplementary material. The senior management will select a handful of presentations and have technical experts review the kernels to validate the calculations and assumptions. These kernel(s) should not be a record of your exploration but only the justification for the statements in the presentation. Kernels in your submission will automatically turn public at the contest deadline, but feel free to share them early and often. Instead of one big kernel, consider having multiple small kernels each tackling a specific topic. Make sure to use Kaggle's new feature that allows users to use output of one kernel as input to the next kernel.\n\nFor the judging criteria, I would add something like\n\nClarity/Concise -- Are the slides professional quality? Do they stand alone without narration? Are there any weak slides that should have been removed? Are the ideas clear and unambiguous?\n\nAnyway, that's my 2 cents. I'm not intending to be a sour grape. I congratulate all the winners. They did a fantastic job. With a little guidance, I think we can collective raise the bar for the next competition.", "votes": 5, "replies": [ { "id": 460072, "postDate": "2019-01-22T23:41:01.500Z", "content": "
\nSummary Slides: A summary of the proposed rule change presented in slide format and delivered as either a PDF or PPT file.\n
\nKernels Analysis: At least one kernel containing the analysis to support your proposal. All kernels submitted must be made public on or before the submission deadline to be eligible. If submitting as a team, all team members must be listed as collaborators on all kernels submitted.\n
\nSubmissions will be judged by the NFL on the following criteria:\n
\nSolution Efficacy - Have you clearly demonstrated, through your data analysis, that you have an understanding of what play features may be associated with concussions and how your proposed rule change(s) will reduce these injuries? Your kernels should be easy to understand and the analysis should be reproducible.\n
\nGame Integrity - Is your proposal actionable by the NFL? Could they implement your rule change and still maintain the integrity of the game? Have you considered the way your proposed changes to game dynamics could introduce new risks to player safety? Strong submissions will demonstrate an understanding for the game overall.\n
Agree. Think the deleted comment was from @hallayang and was going to come back to it - had a lot of good points and information, useful for future reference. Perhaps will consider a repost in a separate new thread? It is quite helpful to hear from the winners, sort of similar to other competitions where the top teams comment on their approach. Maybe after Atlanta? would be nice to hear feedback from the fab4!
", "rawMarkdown": "Agree. Think the deleted comment was from @hallayang and was going to come back to it - had a lot of good points and information, useful for future reference. Perhaps will consider a repost in a separate new thread? It is quite helpful to hear from the winners, sort of similar to other competitions where the top teams comment on their approach. Maybe after Atlanta? would be nice to hear feedback from the fab4!\n", "votes": 1 } ] }, { "id": 459097, "postDate": "2019-01-21T07:18:26.637Z", "content": "Congrats @jpmiller ! Think you probably nailed what was being looked for and did a great professional job.
\n\nPerhaps more clarity on that from the start, even a template designed notebook would have helped. Most kaggle competitions have a submission format via their sample files or model write up requirements, so would not be out of place. Early clarification on use of external data and a pinned thread for external data as in other competitions would have been good too. Usually it is to keep a level playing field so others can benefit from the same external data. It was unclear whether the focus of this competition was meant to be on the NGS data provided, given the Big Data Bowl going on in a similar timeframe, so yes deciding where to spread efforts was hard. Did wonder about the data on players being anonymised or not, without the videos it could possibly have been and then having ages, weights, seasons played might have added some other dimensions. e.g. Seahawks punter Ryan 6ft 200+lbs taking a hit c.f. rookie Steelers McCrane 5ft+ 165lbs (not in this data but as an example). Since getting that data would have required screen scraping from sites that might not allow it and having seen a video by a doctor from Vanderbilt that mentioned the players association reluctance to helmet sensors and how the sensor data might be used - affects on medical privacy, career advances, etc., it seemed anonymity would have been appropriate.
\n\nGood luck in Atlanta. Since the Bears missed out am jealous but not as much!
", "rawMarkdown": "Congrats @jpmiller ! Think you probably nailed what was being looked for and did a great professional job. \n\nPerhaps more clarity on that from the start, even a template designed notebook would have helped. Most kaggle competitions have a submission format via their sample files or model write up requirements, so would not be out of place. Early clarification on use of external data and a pinned thread for external data as in other competitions would have been good too. Usually it is to keep a level playing field so others can benefit from the same external data. It was unclear whether the focus of this competition was meant to be on the NGS data provided, given the Big Data Bowl going on in a similar timeframe, so yes deciding where to spread efforts was hard. Did wonder about the data on players being anonymised or not, without the videos it could possibly have been and then having ages, weights, seasons played might have added some other dimensions. e.g. Seahawks punter Ryan 6ft 200+lbs taking a hit c.f. rookie Steelers McCrane 5ft+ 165lbs (not in this data but as an example). Since getting that data would have required screen scraping from sites that might not allow it and having seen a video by a doctor from Vanderbilt that mentioned the players association reluctance to helmet sensors and how the sensor data might be used - affects on medical privacy, career advances, etc., it seemed anonymity would have been appropriate. \n\nGood luck in Atlanta. Since the Bears missed out am jealous but not as much! ", "votes": 3, "replies": [ { "id": 459644, "postDate": "2019-01-22T05:46:24.003Z", "content": "Thanks. I also like Da Bears. Ideally we would be seeing Johnny Hekker from the Rams punt to Tyreek Hill from KC, but those dang Patriots always find a way to win!
", "rawMarkdown": "Thanks. I also like Da Bears. Ideally we would be seeing Johnny Hekker from the Rams punt to Tyreek Hill from KC, but those dang Patriots always find a way to win!", "votes": 2 } ] }, { "id": 459014, "postDate": "2019-01-21T02:38:20.580Z", "content": "I liked your approach. I had thought the kernal was supposed to be exploratory and the presentation was to be informative. Bad assumption on my part.
\n\nGreat point on nailing one thing.
\n\nGood luck in atlanta
", "rawMarkdown": "I liked your approach. I had thought the kernal was supposed to be exploratory and the presentation was to be informative. Bad assumption on my part.\n\nGreat point on nailing one thing. \n\nGood luck in atlanta", "votes": 4, "replies": [ { "id": 459643, "postDate": "2019-01-22T05:44:01.257Z", "content": "Thanks, Eric. I thought you contributed greatly to this project. I look forward to seeing more of your work in other competitions!
", "rawMarkdown": "Thanks, Eric. I thought you contributed greatly to this project. I look forward to seeing more of your work in other competitions!", "votes": 2 } ] }, { "id": 459677, "postDate": "2019-01-22T06:57:35.577Z", "content": "Congratulations John! You did a great job, and assume that you are one of the two teams presenting. Is it know yet who the two presenters are?
", "rawMarkdown": "Congratulations John! You did a great job, and assume that you are one of the two teams presenting. Is it know yet who the two presenters are?", "votes": 1, "replies": [ { "id": 460067, "postDate": "2019-01-22T23:19:22.107Z", "content": "Thanks Erik. All four of the announced winners are invited to present in Atlanta. I know that two of the others and I are going, don't know yet about the fourth.
", "rawMarkdown": "Thanks Erik. All four of the announced winners are invited to present in Atlanta. I know that two of the others and I are going, don't know yet about the fourth.", "votes": 2 } ] }, { "id": 460082, "postDate": "2019-01-23T00:24:40.753Z", "content": "Congratulations John, you had a fantastic kernel it definitely deserved one of the prizes.
", "rawMarkdown": "Congratulations John, you had a fantastic kernel it definitely deserved one of the prizes.", "votes": 2 }, { "id": 459282, "postDate": "2019-01-21T13:54:46.097Z", "content": "Thanks for posting this John. I'm impressed and inspired by your thoughtful approach in this competition and it showed in your final result!
", "rawMarkdown": "Thanks for posting this John. I'm impressed and inspired by your thoughtful approach in this competition and it showed in your final result!", "votes": 2 }, { "id": 459455, "postDate": "2019-01-21T19:09:17.950Z", "rawMarkdown": "", "votes": 3, "isDeleted": true } ], "comments": [ { "id": 460005, "author_name": "Eric Freeman", "author_url": "", "post_date": "2019-01-22T19:50:28.373000", "content": "The deleted comment had some great ideas about how to think about this competition. Framing it as presentation to senior level executives was obviously the right way to approach the problem. There were some other helpful tips in there, but I don't recall them.
\n\nRe-reading the submission criteria, I would tweak it for the next analytics competition.
\n\n\nA valid submission will include:\n\n\n
\nSummary Slides: A summary of the proposed rule change presented in slide format and delivered as either a PDF or PPT file.\n
\nKernels Analysis: At least one kernel containing the analysis to support your proposal. All kernels submitted must be made public on or before the submission deadline to be eligible. If submitting as a team, all team members must be listed as collaborators on all kernels submitted.\n
\nSubmissions will be judged by the NFL on the following criteria:\n
\nSolution Efficacy - Have you clearly demonstrated, through your data analysis, that you have an understanding of what play features may be associated with concussions and how your proposed rule change(s) will reduce these injuries? Your kernels should be easy to understand and the analysis should be reproducible.\n
\nGame Integrity - Is your proposal actionable by the NFL? Could they implement your rule change and still maintain the integrity of the game? Have you considered the way your proposed changes to game dynamics could introduce new risks to player safety? Strong submissions will demonstrate an understanding for the game overall.\n
I would rephrase the top part (based on my understanding of the judging)
\n\nPresentation: A 5 to 10 minute presentation of the proposed rule change presented in slide format and delivered as either a PDF or PPT file. The audience is senior non-technical management (VP's, directors, etc). It would be helpful but not required to have an appendix which references a kernel to every graph, calculation, and assumption.
\n\nKernels Analysis: Consider this supplementary material. The senior management will select a handful of presentations and have technical experts review the kernels to validate the calculations and assumptions. These kernel(s) should not be a record of your exploration but only the justification for the statements in the presentation. Kernels in your submission will automatically turn public at the contest deadline, but feel free to share them early and often. Instead of one big kernel, consider having multiple small kernels each tackling a specific topic. Make sure to use Kaggle's new feature that allows users to use output of one kernel as input to the next kernel.
\n\nFor the judging criteria, I would add something like
\n\nClarity/Concise -- Are the slides professional quality? Do they stand alone without narration? Are there any weak slides that should have been removed? Are the ideas clear and unambiguous?
\n\nAnyway, that's my 2 cents. I'm not intending to be a sour grape. I congratulate all the winners. They did a fantastic job. With a little guidance, I think we can collective raise the bar for the next competition.
", "votes": 5, "replies": [ { "id": 460072, "author_name": "something4kag", "author_url": "", "post_date": "2019-01-22T23:41:01.500000", "content": "Agree. Think the deleted comment was from @hallayang and was going to come back to it - had a lot of good points and information, useful for future reference. Perhaps will consider a repost in a separate new thread? It is quite helpful to hear from the winners, sort of similar to other competitions where the top teams comment on their approach. Maybe after Atlanta? would be nice to hear feedback from the fab4!
", "votes": 1, "replies": [] } ] }, { "id": 459097, "author_name": "something4kag", "author_url": "", "post_date": "2019-01-21T07:18:26.637000", "content": "Congrats @jpmiller ! Think you probably nailed what was being looked for and did a great professional job.
\n\nPerhaps more clarity on that from the start, even a template designed notebook would have helped. Most kaggle competitions have a submission format via their sample files or model write up requirements, so would not be out of place. Early clarification on use of external data and a pinned thread for external data as in other competitions would have been good too. Usually it is to keep a level playing field so others can benefit from the same external data. It was unclear whether the focus of this competition was meant to be on the NGS data provided, given the Big Data Bowl going on in a similar timeframe, so yes deciding where to spread efforts was hard. Did wonder about the data on players being anonymised or not, without the videos it could possibly have been and then having ages, weights, seasons played might have added some other dimensions. e.g. Seahawks punter Ryan 6ft 200+lbs taking a hit c.f. rookie Steelers McCrane 5ft+ 165lbs (not in this data but as an example). Since getting that data would have required screen scraping from sites that might not allow it and having seen a video by a doctor from Vanderbilt that mentioned the players association reluctance to helmet sensors and how the sensor data might be used - affects on medical privacy, career advances, etc., it seemed anonymity would have been appropriate.
\n\nGood luck in Atlanta. Since the Bears missed out am jealous but not as much!
", "votes": 3, "replies": [ { "id": 459644, "author_name": "JohnM", "author_url": "", "post_date": "2019-01-22T05:46:24.003000", "content": "Thanks. I also like Da Bears. Ideally we would be seeing Johnny Hekker from the Rams punt to Tyreek Hill from KC, but those dang Patriots always find a way to win!
", "votes": 2, "replies": [] } ] }, { "id": 459014, "author_name": "Eric Freeman", "author_url": "", "post_date": "2019-01-21T02:38:20.580000", "content": "I liked your approach. I had thought the kernal was supposed to be exploratory and the presentation was to be informative. Bad assumption on my part.
\n\nGreat point on nailing one thing.
\n\nGood luck in atlanta
", "votes": 4, "replies": [ { "id": 459643, "author_name": "JohnM", "author_url": "", "post_date": "2019-01-22T05:44:01.257000", "content": "Thanks, Eric. I thought you contributed greatly to this project. I look forward to seeing more of your work in other competitions!
", "votes": 2, "replies": [] } ] }, { "id": 459677, "author_name": "Erik Bruin", "author_url": "", "post_date": "2019-01-22T06:57:35.577000", "content": "Congratulations John! You did a great job, and assume that you are one of the two teams presenting. Is it know yet who the two presenters are?
", "votes": 1, "replies": [ { "id": 460067, "author_name": "JohnM", "author_url": "", "post_date": "2019-01-22T23:19:22.107000", "content": "Thanks Erik. All four of the announced winners are invited to present in Atlanta. I know that two of the others and I are going, don't know yet about the fourth.
", "votes": 2, "replies": [] } ] }, { "id": 460082, "author_name": "Garry", "author_url": "", "post_date": "2019-01-23T00:24:40.753000", "content": "Congratulations John, you had a fantastic kernel it definitely deserved one of the prizes.
", "votes": 2, "replies": [] }, { "id": 459282, "author_name": "Rob Mulla", "author_url": "", "post_date": "2019-01-21T13:54:46.097000", "content": "Thanks for posting this John. I'm impressed and inspired by your thoughtful approach in this competition and it showed in your final result!
", "votes": 2, "replies": [] }, { "id": 459455, "author_name": "", "author_url": "", "post_date": "2019-01-21T19:09:17.950000", "content": "", "votes": 3, "replies": [] } ], "raw_markdown_by_id": { "458472": "Thank you everyone for a great competition! I wanted to share a few observations given that this format was a \"first\". Of course, anything I share is merely an opinion that may or may not have helped in this case.\n\n**Information vs Exploration.** I used an informational approach for my main kernel since this was a report. To me that means:\n\n - Bottom line/summary up front\n - Clear outline for the content\n - Conclusions and relevant evidence only\n - Plots with a specific message\n\n**Targeted Analysis.** This was a great case of a lot of data and very little time. I found that a logic tree and some initial exploration helped focus where I wanted to work. It's generally better to prioritize on one thing and kill it vs. spreading your effort across too many opportunities.\n\n**Functional Style.** I spent a lot of time learning how to make my notebook look better. Not just for flair, but to help guide the reader. Section headings, colored fonts, anything that might enhance readability. R users have an advantage with R Markdown reports that are relatively easy to style. Regular notebooks require more manual HTML.\n\nIt would great to hear any observations/challenges/etc that others might have!", "460005": "The deleted comment had some great ideas about how to think about this competition. Framing it as presentation to senior level executives was obviously the right way to approach the problem. There were some other helpful tips in there, but I don't recall them.\n\nRe-reading the submission criteria, I would tweak it for the next analytics competition.\n\n\nA valid submission will include:\n\n\n\nI would rephrase the top part (based on my understanding of the judging)\n\nPresentation: A 5 to 10 minute presentation of the proposed rule change presented in slide format and delivered as either a PDF or PPT file. The audience is senior non-technical management (VP's, directors, etc). It would be helpful but not required to have an appendix which references a kernel to every graph, calculation, and assumption.\n\nKernels Analysis: Consider this supplementary material. The senior management will select a handful of presentations and have technical experts review the kernels to validate the calculations and assumptions. These kernel(s) should not be a record of your exploration but only the justification for the statements in the presentation. Kernels in your submission will automatically turn public at the contest deadline, but feel free to share them early and often. Instead of one big kernel, consider having multiple small kernels each tackling a specific topic. Make sure to use Kaggle's new feature that allows users to use output of one kernel as input to the next kernel.\n\nFor the judging criteria, I would add something like\n\nClarity/Concise -- Are the slides professional quality? Do they stand alone without narration? Are there any weak slides that should have been removed? Are the ideas clear and unambiguous?\n\nAnyway, that's my 2 cents. I'm not intending to be a sour grape. I congratulate all the winners. They did a fantastic job. With a little guidance, I think we can collective raise the bar for the next competition.", "459097": "Congrats @jpmiller ! Think you probably nailed what was being looked for and did a great professional job. \n\nPerhaps more clarity on that from the start, even a template designed notebook would have helped. Most kaggle competitions have a submission format via their sample files or model write up requirements, so would not be out of place. Early clarification on use of external data and a pinned thread for external data as in other competitions would have been good too. Usually it is to keep a level playing field so others can benefit from the same external data. It was unclear whether the focus of this competition was meant to be on the NGS data provided, given the Big Data Bowl going on in a similar timeframe, so yes deciding where to spread efforts was hard. Did wonder about the data on players being anonymised or not, without the videos it could possibly have been and then having ages, weights, seasons played might have added some other dimensions. e.g. Seahawks punter Ryan 6ft 200+lbs taking a hit c.f. rookie Steelers McCrane 5ft+ 165lbs (not in this data but as an example). Since getting that data would have required screen scraping from sites that might not allow it and having seen a video by a doctor from Vanderbilt that mentioned the players association reluctance to helmet sensors and how the sensor data might be used - affects on medical privacy, career advances, etc., it seemed anonymity would have been appropriate. \n\nGood luck in Atlanta. Since the Bears missed out am jealous but not as much! ", "459014": "I liked your approach. I had thought the kernal was supposed to be exploratory and the presentation was to be informative. Bad assumption on my part.\n\nGreat point on nailing one thing. \n\nGood luck in atlanta", "459677": "Congratulations John! You did a great job, and assume that you are one of the two teams presenting. Is it know yet who the two presenters are?", "460082": "Congratulations John, you had a fantastic kernel it definitely deserved one of the prizes.", "459282": "Thanks for posting this John. I'm impressed and inspired by your thoughtful approach in this competition and it showed in your final result!", "459455": "" } }
\nSummary Slides: A summary of the proposed rule change presented in slide format and delivered as either a PDF or PPT file.\n
\nKernels Analysis: At least one kernel containing the analysis to support your proposal. All kernels submitted must be made public on or before the submission deadline to be eligible. If submitting as a team, all team members must be listed as collaborators on all kernels submitted.\n
\nSubmissions will be judged by the NFL on the following criteria:\n
\nSolution Efficacy - Have you clearly demonstrated, through your data analysis, that you have an understanding of what play features may be associated with concussions and how your proposed rule change(s) will reduce these injuries? Your kernels should be easy to understand and the analysis should be reproducible.\n
\nGame Integrity - Is your proposal actionable by the NFL? Could they implement your rule change and still maintain the integrity of the game? Have you considered the way your proposed changes to game dynamics could introduce new risks to player safety? Strong submissions will demonstrate an understanding for the game overall.\n