With respect to public notebook:
\nhttps://www.kaggle.com/code/extraxim/ensemble-csv-for-more-files-lb-25-6
\nwhich scores 25.6 ( In the medal zone), I observe the code just copies the submission file:
*import pandas as pd
\nsub = pd.read_csv(\"/kaggle/input/public-25-6/submission_median.csv\")
\nsub.to_csv(\"submission.csv\", index=False)*
This is surprising as the competition will require the code to take in the test file and
\ngenerate the predictions. However, during the private LB testing, whether such a strategy
\nof just submitting the precomputed csv file will work? I observe the test data of public LB and private LB will be different.
With respect to public notebook:
\nhttps://www.kaggle.com/code/extraxim/ensemble-csv-for-more-files-lb-25-6
\nwhich scores 25.6 ( In the medal zone), I observe the code just copies the submission file:
*import pandas as pd
\nsub = pd.read_csv(\"/kaggle/input/public-25-6/submission_median.csv\")
\nsub.to_csv(\"submission.csv\", index=False)*
This is surprising as the competition will require the code to take in the test file and
\ngenerate the predictions. However, during the private LB testing, whether such a strategy
\nof just submitting the precomputed csv file will work? I observe the test data of public LB and private LB will be different.
I guess the submitted CSV file contains all the test data, and the public LB is calculated based on 25% of it, so during the private LB testing, doesn't need to receive the new test file, just use the other 75%.
", "rawMarkdown": "I guess the submitted CSV file contains all the test data, and the public LB is calculated based on 25% of it, so during the private LB testing, doesn't need to receive the new test file, just use the other 75%.", "votes": null }, { "id": "3231937", "postDate": "06/25/2025 06:38:08", "content": "This is not a code competition. You can upload a csv file as submission.
", "rawMarkdown": "This is not a code competition. You can upload a csv file as submission.", "votes": null }, { "id": "3232437", "postDate": "06/25/2025 17:53:58", "content": "Copying precomputed CSV won't work for private LB—test data differs. Code must process new inputs dynamically to succeed.
", "rawMarkdown": "Copying precomputed CSV won't work for private LB—test data differs. Code must process new inputs dynamically to succeed.", "votes": null }, { "id": "3232451", "postDate": "06/25/2025 18:20:00", "content": "CSV include private test.
", "rawMarkdown": "CSV include private test.", "votes": null }, { "id": "3232574", "postDate": "06/25/2025 23:36:02", "content": "I am new to Kaggle. I have heard that ensemble learning methods that exceed practical computational capabilities often achieve high rankings in Kaggle CSV submission contests. Is it likely that multiple ensemble methods that consume large amounts of resources will rank highly in this contest as well?
", "rawMarkdown": "I am new to Kaggle. I have heard that ensemble learning methods that exceed practical computational capabilities often achieve high rankings in Kaggle CSV submission contests. Is it likely that multiple ensemble methods that consume large amounts of resources will rank highly in this contest as well?", "votes": null }, { "id": "3234211", "postDate": "06/27/2025 16:08:51", "content": "I cross checked the competition website. No where it is mentioned. Can you discuss your source for this input on test data ?
", "rawMarkdown": "I cross checked the competition website. No where it is mentioned. Can you discuss your source for this input on test data ?", "votes": null }, { "id": "3234437", "postDate": "06/27/2025 20:24:42", "content": "This is how every non code competitions works, except for some forecasting competitions. When the public LB score is computed, the private LB score is also computed and kept secret till the end of competition.
", "rawMarkdown": "This is how every non code competitions works, except for some forecasting competitions. When the public LB score is computed, the private LB score is also computed and kept secret till the end of competition.", "votes": null } ], "comments": [ { "id": 3231933, "author_name": "w007425y", "author_url": "", "post_date": "06/25/2025 06:34:52", "content": "I guess the submitted CSV file contains all the test data, and the public LB is calculated based on 25% of it, so during the private LB testing, doesn't need to receive the new test file, just use the other 75%.
", "votes": null, "replies": [ { "id": 3234211, "author_name": "crsuthikshnkumar", "author_url": "", "post_date": "06/27/2025 16:08:51", "content": "I cross checked the competition website. No where it is mentioned. Can you discuss your source for this input on test data ?
", "votes": null, "replies": [ { "id": 3234437, "author_name": "cpmpml", "author_url": "", "post_date": "06/27/2025 20:24:42", "content": "This is how every non code competitions works, except for some forecasting competitions. When the public LB score is computed, the private LB score is also computed and kept secret till the end of competition.
", "votes": null, "replies": [] } ] } ] }, { "id": 3231937, "author_name": "cpmpml", "author_url": "", "post_date": "06/25/2025 06:38:08", "content": "This is not a code competition. You can upload a csv file as submission.
", "votes": null, "replies": [ { "id": 3232574, "author_name": "shunsukehayashi1993", "author_url": "", "post_date": "06/25/2025 23:36:02", "content": "I am new to Kaggle. I have heard that ensemble learning methods that exceed practical computational capabilities often achieve high rankings in Kaggle CSV submission contests. Is it likely that multiple ensemble methods that consume large amounts of resources will rank highly in this contest as well?
", "votes": null, "replies": [] } ] }, { "id": 3232437, "author_name": "pratyushpuri", "author_url": "", "post_date": "06/25/2025 17:53:58", "content": "Copying precomputed CSV won't work for private LB—test data differs. Code must process new inputs dynamically to succeed.
", "votes": null, "replies": [ { "id": 3232451, "author_name": "shlomoron", "author_url": "", "post_date": "06/25/2025 18:20:00", "content": "CSV include private test.
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "3231921": "With respect to public notebook:\nhttps://www.kaggle.com/code/extraxim/ensemble-csv-for-more-files-lb-25-6\nwhich scores 25.6 ( In the medal zone), I observe the code just copies the submission file:\n\n*import pandas as pd\n\nsub = pd.read_csv(\"/kaggle/input/public-25-6/submission_median.csv\")\nsub.to_csv(\"submission.csv\", index=False)*\n\nThis is surprising as the competition will require the code to take in the test file and\ngenerate the predictions. However, during the private LB testing, whether such a strategy \nof just submitting the precomputed csv file will work? I observe the test data of public LB and private LB will be different.", "3231933": "I guess the submitted CSV file contains all the test data, and the public LB is calculated based on 25% of it, so during the private LB testing, doesn't need to receive the new test file, just use the other 75%.", "3231937": "This is not a code competition. You can upload a csv file as submission.", "3232437": "Copying precomputed CSV won't work for private LB—test data differs. Code must process new inputs dynamically to succeed.", "3232451": "CSV include private test.", "3232574": "I am new to Kaggle. I have heard that ensemble learning methods that exceed practical computational capabilities often achieve high rankings in Kaggle CSV submission contests. Is it likely that multiple ensemble methods that consume large amounts of resources will rank highly in this contest as well?", "3234211": "I cross checked the competition website. No where it is mentioned. Can you discuss your source for this input on test data ?", "3234437": "This is how every non code competitions works, except for some forecasting competitions. When the public LB score is computed, the private LB score is also computed and kept secret till the end of competition." }, "source": "meta" }