{ "id": 535093, "title": "Calculate your metric faster and more efficiently!", "url": "/competitions/child-mind-institute-problematic-internet-use/discussion/535093", "author_name": "", "post_date": "2024-09-20T08:37:53.679713900Z", "votes": 31, "comment_count": 1, "views": 0, "content": "

Hello all,

\n

I wish to collate a few materials that may help one to understand the evaluation metric better. This is imperative as a starter process for the assignment-

\n

Explanatory materials

\n
    \n
  1. https://www.kaggle.com/code/taikimori/metric-understand-quadratic-weighted-kappa
    2. \n
  2. https://www.kaggle.com/competitions/learning-agency-lab-automated-essay-scoring-2/discussion/495134
    3. \n
  3. https://www.kaggle.com/competitions/petfinder-adoption-prediction/discussion/76106
    4. \n
  4. https://www.kaggle.com/competitions/liverpool-ion-switching/discussion/132260
    5. \n
  5. https://www.kaggle.com/competitions/playground-series-s3e5/discussion/383479
    6. \n
\n

The below references are useful to deviate from the sklearn cohen_kappa_score and build one's own metric more effectively-

\n\n

YouTube references

\n\n

Best regards!

", "messages": [ { "id": "2993784", "postDate": "09/20/2024 08:37:53", "content": "

Hello all,

\n

I wish to collate a few materials that may help one to understand the evaluation metric better. This is imperative as a starter process for the assignment-

\n

Explanatory materials

\n
    \n
  1. https://www.kaggle.com/code/taikimori/metric-understand-quadratic-weighted-kappa
    2. \n
  2. https://www.kaggle.com/competitions/learning-agency-lab-automated-essay-scoring-2/discussion/495134
    3. \n
  3. https://www.kaggle.com/competitions/petfinder-adoption-prediction/discussion/76106
    4. \n
  4. https://www.kaggle.com/competitions/liverpool-ion-switching/discussion/132260
    5. \n
  5. https://www.kaggle.com/competitions/playground-series-s3e5/discussion/383479
    6. \n
\n

The below references are useful to deviate from the sklearn cohen_kappa_score and build one's own metric more effectively-

\n\n

YouTube references

\n\n

Best regards!

", "rawMarkdown": "Hello all,\n\nI wish to collate a few materials that may help one to understand the evaluation metric better. This is imperative as a starter process for the assignment- \n\n## Explanatory materials\n1. https://www.kaggle.com/code/taikimori/metric-understand-quadratic-weighted-kappa\n2. https://www.kaggle.com/competitions/learning-agency-lab-automated-essay-scoring-2/discussion/495134\n3. https://www.kaggle.com/competitions/petfinder-adoption-prediction/discussion/76106\n4. https://www.kaggle.com/competitions/liverpool-ion-switching/discussion/132260\n5. https://www.kaggle.com/competitions/playground-series-s3e5/discussion/383479\n\nThe below references are useful to deviate from the sklearn cohen_kappa_score and build one's own metric more effectively- \n\n- Kappa metric for PyTorch is as [here](https://github.com/Mamiglia/WeightedKappaLoss) \n- Sklearn methods are not particularly written for performance and speed. Here is a way to expedite the calculation using 3 options [here](https://www.kaggle.com/code/cpmpml/ultra-fast-qwk-calc-method)\n- Given below is another kernel on similar lines, comparing `numpy and numba` effectively [here](https://www.kaggle.com/code/jiweiliu/qwk-cupy-vs-numpy-vs-numba)\n\n## YouTube references \n- https://www.youtube.com/watch?v=VlkTX3_aS1c\n- https://www.youtube.com/watch?v=DfNo32nL_fo\n- https://www.youtube.com/watch?v=K_u3xnBo1Z0\n- https://www.youtube.com/watch?v=zaB6aeu9kNU\n- https://www.youtube.com/watch?v=mCAjVBSgCJI\n\nBest regards!", "votes": null }, { "id": "3008982", "postDate": "10/07/2024 11:13:41", "content": "

very useful information, thank you

", "rawMarkdown": "very useful information, thank you", "votes": null } ], "comments": [ { "id": 3008982, "author_name": "oksanatemruk", "author_url": "", "post_date": "10/07/2024 11:13:41", "content": "

very useful information, thank you

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2993784": "Hello all,\n\nI wish to collate a few materials that may help one to understand the evaluation metric better. This is imperative as a starter process for the assignment- \n\n## Explanatory materials\n1. https://www.kaggle.com/code/taikimori/metric-understand-quadratic-weighted-kappa\n2. https://www.kaggle.com/competitions/learning-agency-lab-automated-essay-scoring-2/discussion/495134\n3. https://www.kaggle.com/competitions/petfinder-adoption-prediction/discussion/76106\n4. https://www.kaggle.com/competitions/liverpool-ion-switching/discussion/132260\n5. https://www.kaggle.com/competitions/playground-series-s3e5/discussion/383479\n\nThe below references are useful to deviate from the sklearn cohen_kappa_score and build one's own metric more effectively- \n\n- Kappa metric for PyTorch is as [here](https://github.com/Mamiglia/WeightedKappaLoss) \n- Sklearn methods are not particularly written for performance and speed. Here is a way to expedite the calculation using 3 options [here](https://www.kaggle.com/code/cpmpml/ultra-fast-qwk-calc-method)\n- Given below is another kernel on similar lines, comparing `numpy and numba` effectively [here](https://www.kaggle.com/code/jiweiliu/qwk-cupy-vs-numpy-vs-numba)\n\n## YouTube references \n- https://www.youtube.com/watch?v=VlkTX3_aS1c\n- https://www.youtube.com/watch?v=DfNo32nL_fo\n- https://www.youtube.com/watch?v=K_u3xnBo1Z0\n- https://www.youtube.com/watch?v=zaB6aeu9kNU\n- https://www.youtube.com/watch?v=mCAjVBSgCJI\n\nBest regards!", "3008982": "very useful information, thank you" }, "source": "meta" }