{ "id": 505767, "title": "I have a question about metric. Why does the host continue to use hacked metric ? ", "url": "/competitions/home-credit-credit-risk-model-stability/discussion/505767", "author_name": "Takumi Mukaiyama", "post_date": "2024-05-19T03:42:29.155000", "votes": 4, "comment_count": 1, "views": 0, "content": "

It is our understanding that Metric's objective is to ensure that the model continues to produce good accuracy. (Does 88 mean after 88 weeks?)
\nAs long as the variance and standard deviation of the predictions are incorporated into the metric, the score will be better if there is an upper and lower limit to the predictions. So, on the currently published Notebook,
\ncondition = y_pred > 0.95
\nis used in the current publicly available Notebook.
\nWhy don't the hosts change to a new metric such as the mean of the weekly Gini coefficient, the mean of the AUC or the F1 value?
\nCan someone please tell me?

\n

Translated with DeepL.com (free version)
\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/475878

", "messages": [ { "id": 2823097, "postDate": "2024-05-19T03:42:29.157Z", "content": "

It is our understanding that Metric's objective is to ensure that the model continues to produce good accuracy. (Does 88 mean after 88 weeks?)
\nAs long as the variance and standard deviation of the predictions are incorporated into the metric, the score will be better if there is an upper and lower limit to the predictions. So, on the currently published Notebook,
\ncondition = y_pred > 0.95
\nis used in the current publicly available Notebook.
\nWhy don't the hosts change to a new metric such as the mean of the weekly Gini coefficient, the mean of the AUC or the F1 value?
\nCan someone please tell me?

\n

Translated with DeepL.com (free version)
\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/475878

", "rawMarkdown": "It is our understanding that Metric's objective is to ensure that the model continues to produce good accuracy. (Does 88 mean after 88 weeks?)\nAs long as the variance and standard deviation of the predictions are incorporated into the metric, the score will be better if there is an upper and lower limit to the predictions. So, on the currently published Notebook,\ncondition = y_pred > 0.95\nis used in the current publicly available Notebook.\nWhy don't the hosts change to a new metric such as the mean of the weekly Gini coefficient, the mean of the AUC or the F1 value?\nCan someone please tell me?\n\n\nTranslated with DeepL.com (free version)\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/475878", "votes": 4 }, { "id": 2823215, "postDate": "2024-05-19T04:52:58.640Z", "rawMarkdown": "", "votes": -1, "isDeleted": true } ], "comments": [ { "id": 2823215, "author_name": "", "author_url": "", "post_date": "2024-05-19T04:52:58.640000", "content": "", "votes": -1, "replies": [] } ], "raw_markdown_by_id": { "2823097": "It is our understanding that Metric's objective is to ensure that the model continues to produce good accuracy. (Does 88 mean after 88 weeks?)\nAs long as the variance and standard deviation of the predictions are incorporated into the metric, the score will be better if there is an upper and lower limit to the predictions. So, on the currently published Notebook,\ncondition = y_pred > 0.95\nis used in the current publicly available Notebook.\nWhy don't the hosts change to a new metric such as the mean of the weekly Gini coefficient, the mean of the AUC or the F1 value?\nCan someone please tell me?\n\n\nTranslated with DeepL.com (free version)\nhttps://www.kaggle.com/competitions/home-credit-credit-risk-model-stability/discussion/475878", "2823215": "" } }