It's a lose/lose situation for both of us, the organizers and competitors. We wasted 2 weeks only for leaking public data.
\n\nI don't like the reactive approach here. Organizers waiting sites getting leaked one by one and people wasting their time trying to find more leaks. At first I thought, this will last couple days, but it really turned to a web scraping competition.
\n\nI think organizers should add all public data to the training set before it's too late.
", "messages": [ { "id": "678187", "postDate": "11/21/2019 04:36:38", "content": "It's a lose/lose situation for both of us, the organizers and competitors. We wasted 2 weeks only for leaking public data.
\n\nI don't like the reactive approach here. Organizers waiting sites getting leaked one by one and people wasting their time trying to find more leaks. At first I thought, this will last couple days, but it really turned to a web scraping competition.
\n\nI think organizers should add all public data to the training set before it's too late.
", "rawMarkdown": "It's a lose/lose situation for both of us, the organizers and competitors. We wasted 2 weeks only for leaking public data.\n\nI don't like the reactive approach here. Organizers waiting sites getting leaked one by one and people wasting their time trying to find more leaks. At first I thought, this will last couple days, but it really turned to a web scraping competition.\n\nI think organizers should add all public data to the training set before it's too late.", "votes": null }, { "id": "678285", "postDate": "11/21/2019 07:28:02", "content": "I second your request.\nAnd I agree the attitude of the organizers, in particular Kaggle staff, has been really disappointing so far.
", "rawMarkdown": "I second your request.\nAnd I agree the attitude of the organizers, in particular Kaggle staff, has been really disappointing so far.", "votes": null }, { "id": "678301", "postDate": "11/21/2019 07:55:48", "content": "Not sure organizers can. @sohier wrote they \"believe that enough of the sites are unavailable online\". He is not sure so I guess all or almost all are available...
\n\nEdited to try to be more constructive.
", "rawMarkdown": "Not sure organizers can. @sohier wrote they \"believe that enough of the sites are unavailable online\". He is not sure so I guess all or almost all are available...\n\n*Edited to try to be more constructive.*", "votes": null }, { "id": "678307", "postDate": "11/21/2019 08:05:49", "content": "1
", "rawMarkdown": "1", "votes": null }, { "id": "678310", "postDate": "11/21/2019 08:06:50", "content": "I'm very consistent with you. But some bad things is happened. I think I have find a site .maybe it's in the Private LB. I decided to open source the data this weekend, Because I want to confirm the correctness of the data. The open source data found on the website may be a little different from the data in the competition.
", "rawMarkdown": "I'm very consistent with you. But some bad things is happened. I think I have find a site .maybe it's in the Private LB. I decided to open source the data this weekend, Because I want to confirm the correctness of the data. The open source data found on the website may be a little different from the data in the competition.", "votes": null }, { "id": "678315", "postDate": "11/21/2019 08:18:30", "content": "As much as I agree with this, I don't think organizers would fulfill this request. We still have 4 weeks find all the data, disclose it on the external data thread and build the best model we can. Quite frankly, I stopped caring about the leaked data and just focused on modeling.
", "rawMarkdown": "As much as I agree with this, I don't think organizers would fulfill this request. We still have 4 weeks find all the data, disclose it on the external data thread and build the best model we can. Quite frankly, I stopped caring about the leaked data and just focused on modeling.", "votes": null }, { "id": "678318", "postDate": "11/21/2019 08:31:32", "content": "I'm totally agree with you.
\n\nAccording to the description of this competition, winner will be invited to an annual conference of ASHRAE next year, so now I'm really afraid that what winner will say in that event. I hope the winner will be more smarter than me.
", "rawMarkdown": "I'm totally agree with you. \n\nAccording to the description of this competition, winner will be invited to an annual conference of ASHRAE next year, so now I'm really afraid that what winner will say in that event. I hope the winner will be more smarter than me.", "votes": null }, { "id": "678331", "postDate": "11/21/2019 08:47:30", "content": "The more interesting thing - is there anything than is not publicly available?
", "rawMarkdown": "The more interesting thing - is there anything than is **not** publicly available?", "votes": null }, { "id": "678372", "postDate": "11/21/2019 09:38:30", "content": "I want to be far from top LB. stop working on improvement aww....
", "rawMarkdown": "I want to be far from top LB. stop working on improvement aww....", "votes": null }, { "id": "678378", "postDate": "11/21/2019 09:48:42", "content": "I agree with @pdnartreb : I don't think the organizers are certain of what data is publicly available. My guess is they will remove the publicly available data from private LB at the end of the competition based on what is shared in the competition.
", "rawMarkdown": "I agree with @pdnartreb : I don't think the organizers are certain of what data is publicly available. My guess is they will remove the publicly available data from private LB at the end of the competition based on what is shared in the competition.", "votes": null }, { "id": "678380", "postDate": "11/21/2019 09:54:48", "content": "So who scrapes the most and shares the least will win this competition
", "rawMarkdown": "So who scrapes the most and shares the least will win this competition", "votes": null }, { "id": "678392", "postDate": "11/21/2019 10:09:20", "content": "I think a week is enough time to adjust your models with additional training data provided others share their scraped data within the final week if it comes to that. So, that edge can't be that large.
", "rawMarkdown": "I think a week is enough time to adjust your models with additional training data provided others share their scraped data within the final week if it comes to that. So, that edge can't be that large.", "votes": null }, { "id": "678400", "postDate": "11/21/2019 10:24:50", "content": "Hi everyone, congratulations to all the fantastic Kagglers on here! Apologies for the newbie question but this is my first competition.
\n\nI don't understand what additional information do these data leaks exactly bring to the models? Whilst I understand that these leaks may help to boost public LB score would they still be useful on private LB if the site_ids are different? In fact would these leaks not be bad for generalising models and cause overfitting to train set?
", "rawMarkdown": "Hi everyone, congratulations to all the fantastic Kagglers on here! Apologies for the newbie question but this is my first competition. \n\nI don't understand what additional information do these data leaks exactly bring to the models? Whilst I understand that these leaks may help to boost public LB score would they still be useful on private LB if the site_ids are different? In fact would these leaks not be bad for generalising models and cause overfitting to train set?", "votes": null }, { "id": "678507", "postDate": "11/21/2019 13:32:19", "content": "What if someone has scraped and hasn't shared the same here , one can simply create a submission with that data and get a bronze easily. Who / How can it be ensured that All data which is publicly available is listed in the discussion forums?
", "rawMarkdown": "What if someone has scraped and hasn't shared the same here , one can simply create a submission with that data and get a bronze easily. Who / How can it be ensured that ***All*** data which is publicly available is listed in the discussion forums?", "votes": null }, { "id": "678641", "postDate": "11/21/2019 17:22:20", "content": "Thanks everyone for all the discussion and input around this concern. We are not planning to release any further data. The reason is that, while there are edge cases where the data could in theory be located, it is unlikely. Releasing data prematurely would unnecessarily degrade the utility of the public leaderboard. While also quite unlikely, it is technically possible that one of our data partners has published data without our knowledge since we last spoke to them, which would render a publication effort pointless.
\n\nThere is still a range of scores on the leaderboard which would not be the case if it was determined solely by public data, so there is still an opportunity to assess your models there. As noted here, we will ensure that any meter readings that can be publicly downloaded are not included in the private set regardless.
", "rawMarkdown": "Thanks everyone for all the discussion and input around this concern. We are not planning to release any further data. The reason is that, while there are edge cases where the data could in theory be located, it is unlikely. Releasing data prematurely would unnecessarily degrade the utility of the public leaderboard. While also quite unlikely, it is technically possible that one of our data partners has published data without our knowledge since we last spoke to them, which would render a publication effort pointless.\n\nThere is still a range of scores on the leaderboard which would not be the case if it was determined solely by public data, so there is still an opportunity to assess your models there. [As noted here](https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357#latest-678477), we will ensure that any meter readings that can be publicly downloaded are not included in the private set regardless.", "votes": null }, { "id": "678697", "postDate": "11/21/2019 18:27:26", "content": "I think I know what you meant by edge cases. Site 4 Berkeley public data is very close to competition data, but it's not same. There is no direct advantage of using it. I'm pretty sure there are lots of smart people who are capable of reverse engineering the transformation. It looks like you are still underestimating competitors here.
", "rawMarkdown": "I think I know what you meant by edge cases. Site 4 Berkeley public data is very close to competition data, but it's not same. There is no direct advantage of using it. I'm pretty sure there are lots of smart people who are capable of reverse engineering the transformation. It looks like you are still underestimating competitors here.", "votes": null }, { "id": "678715", "postDate": "11/21/2019 18:46:40", "content": "That is not what I was referring to.
", "rawMarkdown": "That is not what I was referring to.", "votes": null }, { "id": "679054", "postDate": "11/22/2019 07:57:02", "content": "@sohier What we/I can/could hear… The pipeline to win this competition is: scrap the more, probably you will have to perform some kind of (black hat in this case) hacking to find all the data (I guess these are the edge cases you are referring to and I could illustrate that), don't share and pray that no one find/share them, build a model (I already have feature engineering ideas) that overfits the 2018 (private) data you have downloaded but don't use them explicitly…
\nSharing is from my point of view the biggest asset of Kaggle.
Edited to try to be more constructive.
", "rawMarkdown": "sohier What we/I can/could hear… The pipeline to win this competition is: scrap the more, probably you will have to perform some kind of (black hat in this case) hacking to find all the data (I guess these are the edge cases you are referring to and I could illustrate that), don't share and pray that no one find/share them, build a model (I already have feature engineering ideas) that overfits the 2018 (private) data you have downloaded but don't use them explicitly…\nSharing is from my point of view the biggest asset of Kaggle.\n\n*Edited to try to be more constructive.*", "votes": null }, { "id": "679073", "postDate": "11/22/2019 08:49:09", "content": "Aaaaand here goes Google Analytics v2.0. Most stable plan atm - wait for entry deadline, get all data and try to predict something. Cheaters plan - scrap some more data, hide it and use indirectly for validating.
", "rawMarkdown": "Aaaaand here goes Google Analytics v2.0. Most stable plan atm - wait for entry deadline, get all data and try to predict something. Cheaters plan - scrap some more data, hide it and use indirectly for validating.", "votes": null }, { "id": "679230", "postDate": "11/22/2019 13:24:58", "content": "@claytonmiller @chrisbalbach, could you please give your insight regarding data leakage issue? Is it acceptable to your team to receive solutions with high percentage model generalization based on web scraping and/or leakage result?
", "rawMarkdown": "claytonmiller @chrisbalbach, could you please give your insight regarding data leakage issue? Is it acceptable to your team to receive solutions with high percentage model generalization based on web scraping and/or leakage result?", "votes": null }, { "id": "679430", "postDate": "11/22/2019 19:15:56", "content": "☝️ Would very much like to know this too!
", "rawMarkdown": "☝️ Would very much like to know this too!", "votes": null }, { "id": "679728", "postDate": "11/23/2019 09:37:41", "content": "I apologize for the words that I used. Reading once more what I wrote showed me that they were not translating my thoughts, not relevant and not constructive so I edited my posts. I believe that organizers are doing their best to purpose an interesting competition. Frustration and disappointment are not easy to deal with…
", "rawMarkdown": "I apologize for the words that I used. Reading once more what I wrote showed me that they were not translating my thoughts, not relevant and not constructive so I edited my posts. I believe that organizers are doing their best to purpose an interesting competition. Frustration and disappointment are not easy to deal with...", "votes": null }, { "id": "679733", "postDate": "11/23/2019 09:53:02", "content": ":-)
", "rawMarkdown": ":-)", "votes": null }, { "id": "717328", "postDate": "01/13/2020 03:19:35", "content": "Hi @pdnartreb and @projdev -- thanks for your patience during the competition. We understood it was a bit frustrating, but I think the Kaggle team did a good job of making sure the Private Leaderboard kept its integrity. We definitely realized that there are improvements for the next competitions we may try to host. The end goal was to get the building science and data science communities to understand each other better, and I think we at least accomplished that. I stayed quiet during the competition so as to not throw a wrench in the discussion inadvertently.
", "rawMarkdown": "Hi @pdnartreb and @projdev -- thanks for your patience during the competition. We understood it was a bit frustrating, but I think the Kaggle team did a good job of making sure the Private Leaderboard kept its integrity. We definitely realized that there are improvements for the next competitions we may try to host. The end goal was to get the building science and data science communities to understand each other better, and I think we at least accomplished that. I stayed quiet during the competition so as to not throw a wrench in the discussion inadvertently.", "votes": null }, { "id": "1400326", "postDate": "07/26/2021 07:58:51", "content": "great my friend nice work
", "rawMarkdown": "great my friend nice work", "votes": null } ], "comments": [ { "id": 1400326, "author_name": "mohamedsarwat", "author_url": "", "post_date": "07/26/2021 07:58:51", "content": "great my friend nice work
", "votes": null, "replies": [] }, { "id": 678285, "author_name": "stecasasso", "author_url": "", "post_date": "11/21/2019 07:28:02", "content": "I second your request.\nAnd I agree the attitude of the organizers, in particular Kaggle staff, has been really disappointing so far.
", "votes": null, "replies": [] }, { "id": 678301, "author_name": "pdnartreb", "author_url": "", "post_date": "11/21/2019 07:55:48", "content": "Not sure organizers can. @sohier wrote they \"believe that enough of the sites are unavailable online\". He is not sure so I guess all or almost all are available...
\n\nEdited to try to be more constructive.
", "votes": null, "replies": [] }, { "id": 678307, "author_name": "whatpp", "author_url": "", "post_date": "11/21/2019 08:05:49", "content": "1
", "votes": null, "replies": [] }, { "id": 678310, "author_name": "pp2file", "author_url": "", "post_date": "11/21/2019 08:06:50", "content": "I'm very consistent with you. But some bad things is happened. I think I have find a site .maybe it's in the Private LB. I decided to open source the data this weekend, Because I want to confirm the correctness of the data. The open source data found on the website may be a little different from the data in the competition.
", "votes": null, "replies": [] }, { "id": 678315, "author_name": "teeyee314", "author_url": "", "post_date": "11/21/2019 08:18:30", "content": "As much as I agree with this, I don't think organizers would fulfill this request. We still have 4 weeks find all the data, disclose it on the external data thread and build the best model we can. Quite frankly, I stopped caring about the leaked data and just focused on modeling.
", "votes": null, "replies": [] }, { "id": 678318, "author_name": "yamsam", "author_url": "", "post_date": "11/21/2019 08:31:32", "content": "I'm totally agree with you.
\n\nAccording to the description of this competition, winner will be invited to an annual conference of ASHRAE next year, so now I'm really afraid that what winner will say in that event. I hope the winner will be more smarter than me.
", "votes": null, "replies": [] }, { "id": 678331, "author_name": "nroman", "author_url": "", "post_date": "11/21/2019 08:47:30", "content": "The more interesting thing - is there anything than is not publicly available?
", "votes": null, "replies": [ { "id": 679430, "author_name": "carlolepelaars", "author_url": "", "post_date": "11/22/2019 19:15:56", "content": "☝️ Would very much like to know this too!
", "votes": null, "replies": [] } ] }, { "id": 678372, "author_name": "hanjoonchoe", "author_url": "", "post_date": "11/21/2019 09:38:30", "content": "I want to be far from top LB. stop working on improvement aww....
", "votes": null, "replies": [] }, { "id": 678378, "author_name": "rohanrao", "author_url": "", "post_date": "11/21/2019 09:48:42", "content": "I agree with @pdnartreb : I don't think the organizers are certain of what data is publicly available. My guess is they will remove the publicly available data from private LB at the end of the competition based on what is shared in the competition.
", "votes": null, "replies": [ { "id": 678380, "author_name": "gunesevitan", "author_url": "", "post_date": "11/21/2019 09:54:48", "content": "So who scrapes the most and shares the least will win this competition
", "votes": null, "replies": [] }, { "id": 678392, "author_name": "teeyee314", "author_url": "", "post_date": "11/21/2019 10:09:20", "content": "I think a week is enough time to adjust your models with additional training data provided others share their scraped data within the final week if it comes to that. So, that edge can't be that large.
", "votes": null, "replies": [] }, { "id": 678507, "author_name": "avikrams", "author_url": "", "post_date": "11/21/2019 13:32:19", "content": "What if someone has scraped and hasn't shared the same here , one can simply create a submission with that data and get a bronze easily. Who / How can it be ensured that All data which is publicly available is listed in the discussion forums?
", "votes": null, "replies": [] } ] }, { "id": 678400, "author_name": "spenot09", "author_url": "", "post_date": "11/21/2019 10:24:50", "content": "Hi everyone, congratulations to all the fantastic Kagglers on here! Apologies for the newbie question but this is my first competition.
\n\nI don't understand what additional information do these data leaks exactly bring to the models? Whilst I understand that these leaks may help to boost public LB score would they still be useful on private LB if the site_ids are different? In fact would these leaks not be bad for generalising models and cause overfitting to train set?
", "votes": null, "replies": [] }, { "id": 678641, "author_name": "sohier", "author_url": "", "post_date": "11/21/2019 17:22:20", "content": "Thanks everyone for all the discussion and input around this concern. We are not planning to release any further data. The reason is that, while there are edge cases where the data could in theory be located, it is unlikely. Releasing data prematurely would unnecessarily degrade the utility of the public leaderboard. While also quite unlikely, it is technically possible that one of our data partners has published data without our knowledge since we last spoke to them, which would render a publication effort pointless.
\n\nThere is still a range of scores on the leaderboard which would not be the case if it was determined solely by public data, so there is still an opportunity to assess your models there. As noted here, we will ensure that any meter readings that can be publicly downloaded are not included in the private set regardless.
", "votes": null, "replies": [ { "id": 678697, "author_name": "gunesevitan", "author_url": "", "post_date": "11/21/2019 18:27:26", "content": "I think I know what you meant by edge cases. Site 4 Berkeley public data is very close to competition data, but it's not same. There is no direct advantage of using it. I'm pretty sure there are lots of smart people who are capable of reverse engineering the transformation. It looks like you are still underestimating competitors here.
", "votes": null, "replies": [] }, { "id": 678715, "author_name": "sohier", "author_url": "", "post_date": "11/21/2019 18:46:40", "content": "That is not what I was referring to.
", "votes": null, "replies": [] }, { "id": 679054, "author_name": "pdnartreb", "author_url": "", "post_date": "11/22/2019 07:57:02", "content": "@sohier What we/I can/could hear… The pipeline to win this competition is: scrap the more, probably you will have to perform some kind of (black hat in this case) hacking to find all the data (I guess these are the edge cases you are referring to and I could illustrate that), don't share and pray that no one find/share them, build a model (I already have feature engineering ideas) that overfits the 2018 (private) data you have downloaded but don't use them explicitly…
\nSharing is from my point of view the biggest asset of Kaggle.
Edited to try to be more constructive.
", "votes": null, "replies": [] }, { "id": 679230, "author_name": "projdev", "author_url": "", "post_date": "11/22/2019 13:24:58", "content": "@claytonmiller @chrisbalbach, could you please give your insight regarding data leakage issue? Is it acceptable to your team to receive solutions with high percentage model generalization based on web scraping and/or leakage result?
", "votes": null, "replies": [] }, { "id": 717328, "author_name": "claytonmiller", "author_url": "", "post_date": "01/13/2020 03:19:35", "content": "Hi @pdnartreb and @projdev -- thanks for your patience during the competition. We understood it was a bit frustrating, but I think the Kaggle team did a good job of making sure the Private Leaderboard kept its integrity. We definitely realized that there are improvements for the next competitions we may try to host. The end goal was to get the building science and data science communities to understand each other better, and I think we at least accomplished that. I stayed quiet during the competition so as to not throw a wrench in the discussion inadvertently.
", "votes": null, "replies": [] } ] }, { "id": 679073, "author_name": "wispzero", "author_url": "", "post_date": "11/22/2019 08:49:09", "content": "Aaaaand here goes Google Analytics v2.0. Most stable plan atm - wait for entry deadline, get all data and try to predict something. Cheaters plan - scrap some more data, hide it and use indirectly for validating.
", "votes": null, "replies": [] }, { "id": 679728, "author_name": "pdnartreb", "author_url": "", "post_date": "11/23/2019 09:37:41", "content": "I apologize for the words that I used. Reading once more what I wrote showed me that they were not translating my thoughts, not relevant and not constructive so I edited my posts. I believe that organizers are doing their best to purpose an interesting competition. Frustration and disappointment are not easy to deal with…
", "votes": null, "replies": [ { "id": 679733, "author_name": "rohanrao", "author_url": "", "post_date": "11/23/2019 09:53:02", "content": ":-)
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "678187": "It's a lose/lose situation for both of us, the organizers and competitors. We wasted 2 weeks only for leaking public data.\n\nI don't like the reactive approach here. Organizers waiting sites getting leaked one by one and people wasting their time trying to find more leaks. At first I thought, this will last couple days, but it really turned to a web scraping competition.\n\nI think organizers should add all public data to the training set before it's too late.", "678285": "I second your request.\nAnd I agree the attitude of the organizers, in particular Kaggle staff, has been really disappointing so far.", "678301": "Not sure organizers can. @sohier wrote they \"believe that enough of the sites are unavailable online\". He is not sure so I guess all or almost all are available...\n\n*Edited to try to be more constructive.*", "678307": "1", "678310": "I'm very consistent with you. But some bad things is happened. I think I have find a site .maybe it's in the Private LB. I decided to open source the data this weekend, Because I want to confirm the correctness of the data. The open source data found on the website may be a little different from the data in the competition.", "678315": "As much as I agree with this, I don't think organizers would fulfill this request. We still have 4 weeks find all the data, disclose it on the external data thread and build the best model we can. Quite frankly, I stopped caring about the leaked data and just focused on modeling.", "678318": "I'm totally agree with you. \n\nAccording to the description of this competition, winner will be invited to an annual conference of ASHRAE next year, so now I'm really afraid that what winner will say in that event. I hope the winner will be more smarter than me.", "678331": "The more interesting thing - is there anything than is **not** publicly available?", "678372": "I want to be far from top LB. stop working on improvement aww....", "678378": "I agree with @pdnartreb : I don't think the organizers are certain of what data is publicly available. My guess is they will remove the publicly available data from private LB at the end of the competition based on what is shared in the competition.", "678380": "So who scrapes the most and shares the least will win this competition", "678392": "I think a week is enough time to adjust your models with additional training data provided others share their scraped data within the final week if it comes to that. So, that edge can't be that large.", "678400": "Hi everyone, congratulations to all the fantastic Kagglers on here! Apologies for the newbie question but this is my first competition. \n\nI don't understand what additional information do these data leaks exactly bring to the models? Whilst I understand that these leaks may help to boost public LB score would they still be useful on private LB if the site_ids are different? In fact would these leaks not be bad for generalising models and cause overfitting to train set?", "678507": "What if someone has scraped and hasn't shared the same here , one can simply create a submission with that data and get a bronze easily. Who / How can it be ensured that ***All*** data which is publicly available is listed in the discussion forums?", "678641": "Thanks everyone for all the discussion and input around this concern. We are not planning to release any further data. The reason is that, while there are edge cases where the data could in theory be located, it is unlikely. Releasing data prematurely would unnecessarily degrade the utility of the public leaderboard. While also quite unlikely, it is technically possible that one of our data partners has published data without our knowledge since we last spoke to them, which would render a publication effort pointless.\n\nThere is still a range of scores on the leaderboard which would not be the case if it was determined solely by public data, so there is still an opportunity to assess your models there. [As noted here](https://www.kaggle.com/c/ashrae-energy-prediction/discussion/117357#latest-678477), we will ensure that any meter readings that can be publicly downloaded are not included in the private set regardless.", "678697": "I think I know what you meant by edge cases. Site 4 Berkeley public data is very close to competition data, but it's not same. There is no direct advantage of using it. I'm pretty sure there are lots of smart people who are capable of reverse engineering the transformation. It looks like you are still underestimating competitors here.", "678715": "That is not what I was referring to.", "679054": "sohier What we/I can/could hear… The pipeline to win this competition is: scrap the more, probably you will have to perform some kind of (black hat in this case) hacking to find all the data (I guess these are the edge cases you are referring to and I could illustrate that), don't share and pray that no one find/share them, build a model (I already have feature engineering ideas) that overfits the 2018 (private) data you have downloaded but don't use them explicitly…\nSharing is from my point of view the biggest asset of Kaggle.\n\n*Edited to try to be more constructive.*", "679073": "Aaaaand here goes Google Analytics v2.0. Most stable plan atm - wait for entry deadline, get all data and try to predict something. Cheaters plan - scrap some more data, hide it and use indirectly for validating.", "679230": "claytonmiller @chrisbalbach, could you please give your insight regarding data leakage issue? Is it acceptable to your team to receive solutions with high percentage model generalization based on web scraping and/or leakage result?", "679430": "☝️ Would very much like to know this too!", "679728": "I apologize for the words that I used. Reading once more what I wrote showed me that they were not translating my thoughts, not relevant and not constructive so I edited my posts. I believe that organizers are doing their best to purpose an interesting competition. Frustration and disappointment are not easy to deal with...", "679733": ":-)", "717328": "Hi @pdnartreb and @projdev -- thanks for your patience during the competition. We understood it was a bit frustrating, but I think the Kaggle team did a good job of making sure the Private Leaderboard kept its integrity. We definitely realized that there are improvements for the next competitions we may try to host. The end goal was to get the building science and data science communities to understand each other better, and I think we at least accomplished that. I stayed quiet during the competition so as to not throw a wrench in the discussion inadvertently.", "1400326": "great my friend nice work" }, "source": "meta" }