{ "id": 556814, "title": "I have two problems confusing me. Could someone please help me?", "url": "/competitions/czii-cryo-et-object-identification/discussion/556814", "author_name": "", "post_date": "2025-01-15T08:17:46.213428Z", "votes": null, "comment_count": 5, "views": 0, "content": "
    \n
  1. What samples are the predictions for in the final submission? Isn't the test set included in the training set?
    2. \n
  2. What are the requirements for the F-beta score? What should we pay more attention to?
    3. \n
", "messages": [ { "id": "3097282", "postDate": "01/15/2025 08:17:46", "content": "
    \n
  1. What samples are the predictions for in the final submission? Isn't the test set included in the training set?
    2. \n
  2. What are the requirements for the F-beta score? What should we pay more attention to?
    3. \n
", "rawMarkdown": "1. What samples are the predictions for in the final submission? Isn't the test set included in the training set?\n2. What are the requirements for the F-beta score? What should we pay more attention to?", "votes": null }, { "id": "3097384", "postDate": "01/15/2025 10:06:08", "content": "
    \n

  1. /kaggle/input/czii-cryo-et-object-identification/test/static/ExperimentRuns
    \nIt's a bit confusing, because there are just 3 samples in there. When you submit your code for scoring the other 497 are populated. They give you 3 just to test your code works when saving, or in interactive mode.

    2. \n

  2. F-beta with a beta of 4 greatly prioritizes recall over precision. Or in more practical terms: false positives are penalized less than false negatives, so you might want to consider using lower thresholds than you otherwise would.

    3. \n
", "rawMarkdown": "1. `/kaggle/input/czii-cryo-et-object-identification/test/static/ExperimentRuns` \nIt's a bit confusing, because there are just 3 samples in there. When you submit your code for scoring the other 497 are populated. They give you 3 just to test your code works when saving, or in interactive mode.\n\n2. F-beta with a beta of 4 greatly prioritizes recall over precision. Or in more practical terms: false positives are penalized less than false negatives, so you might want to consider using lower thresholds than you otherwise would.", "votes": null }, { "id": "3097405", "postDate": "01/15/2025 10:21:27", "content": "

There's a discussion of f-beta here:
\nhttps://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750

", "rawMarkdown": "There's a discussion of f-beta here:\n[https://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750](https://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750)", "votes": null }, { "id": "3097694", "postDate": "01/15/2025 16:01:18", "content": "

\"\"

\n

Hey @davidlist
\nWhat is this? I am seeing it for the first time

", "rawMarkdown": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F15856017%2F41c605d79c66a3eb11199ac6c070f1cc%2FScreenshot%202025-01-15%20212953.png?generation=1736956826699170&alt=media)\n\nHey @davidlist \nWhat is this? I am seeing it for the first time", "votes": null }, { "id": "3097719", "postDate": "01/15/2025 16:29:51", "content": "

First rule of fight club - read all discussions first :-)

", "rawMarkdown": "First rule of fight club - read all discussions first :-)", "votes": null }, { "id": "3097869", "postDate": "01/15/2025 18:55:28", "content": "

@vigneshwar472 It talks about it in the competition overview: https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview.

", "rawMarkdown": "vigneshwar472 It talks about it in the competition overview: [https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview](https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview).", "votes": null } ], "comments": [ { "id": 3097384, "author_name": "ollypowell", "author_url": "", "post_date": "01/15/2025 10:06:08", "content": "
    \n

  1. /kaggle/input/czii-cryo-et-object-identification/test/static/ExperimentRuns
    \nIt's a bit confusing, because there are just 3 samples in there. When you submit your code for scoring the other 497 are populated. They give you 3 just to test your code works when saving, or in interactive mode.

    2. \n

  2. F-beta with a beta of 4 greatly prioritizes recall over precision. Or in more practical terms: false positives are penalized less than false negatives, so you might want to consider using lower thresholds than you otherwise would.

    3. \n
", "votes": null, "replies": [] }, { "id": 3097405, "author_name": "davidlist", "author_url": "", "post_date": "01/15/2025 10:21:27", "content": "

There's a discussion of f-beta here:
\nhttps://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750

", "votes": null, "replies": [ { "id": 3097694, "author_name": "vigneshwar472", "author_url": "", "post_date": "01/15/2025 16:01:18", "content": "

\"\"

\n

Hey @davidlist
\nWhat is this? I am seeing it for the first time

", "votes": null, "replies": [ { "id": 3097869, "author_name": "davidlist", "author_url": "", "post_date": "01/15/2025 18:55:28", "content": "

@vigneshwar472 It talks about it in the competition overview: https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview.

", "votes": null, "replies": [] } ] } ] }, { "id": 3097719, "author_name": "maksimovka", "author_url": "", "post_date": "01/15/2025 16:29:51", "content": "

First rule of fight club - read all discussions first :-)

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "3097282": "1. What samples are the predictions for in the final submission? Isn't the test set included in the training set?\n2. What are the requirements for the F-beta score? What should we pay more attention to?", "3097384": "1. `/kaggle/input/czii-cryo-et-object-identification/test/static/ExperimentRuns` \nIt's a bit confusing, because there are just 3 samples in there. When you submit your code for scoring the other 497 are populated. They give you 3 just to test your code works when saving, or in interactive mode.\n\n2. F-beta with a beta of 4 greatly prioritizes recall over precision. Or in more practical terms: false positives are penalized less than false negatives, so you might want to consider using lower thresholds than you otherwise would.", "3097405": "There's a discussion of f-beta here:\n[https://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750](https://www.kaggle.com/competitions/czii-cryo-et-object-identification/discussion/544750)", "3097694": "![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F15856017%2F41c605d79c66a3eb11199ac6c070f1cc%2FScreenshot%202025-01-15%20212953.png?generation=1736956826699170&alt=media)\n\nHey @davidlist \nWhat is this? I am seeing it for the first time", "3097719": "First rule of fight club - read all discussions first :-)", "3097869": "vigneshwar472 It talks about it in the competition overview: [https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview](https://www.kaggle.com/competitions/czii-cryo-et-object-identification/overview)." }, "source": "meta" }