Hi I saw my selected submission
\nYes I did unshuffle the data and use shift(-lag) features. If this is the reason why, I have nothing to say.
\nBut if it's because I submitted .csv file, I do have a notebook but it's using too much features the memory of kaggle notebook is not enough, I ran it locally. @drwtrading
\nhttps://www.kaggle.com/code/menghaozhao/drw-lgbm
Same here. I also submitted a csv file, but I think it has been invalidated because of using the time structure (lag features which is technically not future peeking). Though I have no idea how they decided about that with just the csv submission and without the code. I'm pretty sure getting scores as high as 0.95 is impossible without using the time, so I'm really curious to see notebooks from those who have not been invalidated yet.
", "rawMarkdown": "Same here. I also submitted a csv file, but I think it has been invalidated because of using the time structure (lag features which is technically not future peeking). Though I have no idea how they decided about that with just the csv submission and without the code. I'm pretty sure getting scores as high as 0.95 is impossible without using the time, so I'm really curious to see notebooks from those who have not been invalidated yet." }, { "id": 3254204, "postDate": "2025-07-26T02:45:04.260Z", "content": "\n", "rawMarkdown": ">This means any direct use of test data order or time-series continuity—whether implicit or reconstructed—is not allowed and will be considered future peeking.", "votes": -1, "replies": [ { "id": 3254223, "postDate": "2025-07-26T03:32:25.873Z", "rawMarkdown": "", "isDeleted": true } ] }, { "id": 3254196, "postDate": "2025-07-26T02:11:45.127Z", "content": "This means any direct use of test data order or time-series continuity—whether implicit or reconstructed—is not allowed and will be considered future peeking.
\n
Hi I saw my selected submission
\nYes I did unshuffle the data and use shift(-lag) features. If this is the reason why, I have nothing to say.
\nBut if it's because I submitted .csv file, I do have a notebook but it's using too much features the memory of kaggle notebook is not enough, I ran it locally. @drwtrading
\nhttps://www.kaggle.com/code/menghaozhao/drw-lgbm
Same here. I also submitted a csv file, but I think it has been invalidated because of using the time structure (lag features which is technically not future peeking). Though I have no idea how they decided about that with just the csv submission and without the code. I'm pretty sure getting scores as high as 0.95 is impossible without using the time, so I'm really curious to see notebooks from those who have not been invalidated yet.
", "votes": 0, "replies": [] }, { "id": 3254204, "author_name": "byunjins", "author_url": "", "post_date": "2025-07-26T02:45:04.260000", "content": "\n", "votes": -1, "replies": [ { "id": 3254223, "author_name": "", "author_url": "", "post_date": "2025-07-26T03:32:25.873000", "content": "", "votes": 0, "replies": [] } ] }, { "id": 3254368, "author_name": "", "author_url": "", "post_date": "2025-07-26T11:13:23.663000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "3254473": "Same here. I also submitted a csv file, but I think it has been invalidated because of using the time structure (lag features which is technically not future peeking). Though I have no idea how they decided about that with just the csv submission and without the code. I'm pretty sure getting scores as high as 0.95 is impossible without using the time, so I'm really curious to see notebooks from those who have not been invalidated yet.", "3254204": ">This means any direct use of test data order or time-series continuity—whether implicit or reconstructed—is not allowed and will be considered future peeking.", "3254196": "Hi I saw my selected submission \nYes I did unshuffle the data and use shift(-lag) features. If this is the reason why, I have nothing to say.\nBut if it's because I submitted .csv file, I do have a notebook but it's using too much features the memory of kaggle notebook is not enough, I ran it locally. @drwtrading \nhttps://www.kaggle.com/code/menghaozhao/drw-lgbm", "3254368": "" } }This means any direct use of test data order or time-series continuity—whether implicit or reconstructed—is not allowed and will be considered future peeking.
\n