![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F1268857%2F1a122d78eeb363235c07d7618227244a%2FIMG_20200828_090500.jpg?generation=1598603516828680&alt=media\")
I am sorry if this is already answered but can someone please explain the confidence part and why it has ml as unit in it(according to the data description) ?????
", "messages": [ { "id": "987662", "postDate": "08/27/2020 12:29:45", "content": "I am sorry if this is already answered but can someone please explain the confidence part and why it has ml as unit in it(according to the data description) ?????
", "rawMarkdown": "I am sorry if this is already answered but can someone please explain the confidence part and why it has ml as unit in it(according to the data description) ?????", "votes": null }, { "id": "987673", "postDate": "08/27/2020 12:35:08", "content": "It's the number of ml you think you might be off regarding your prediction
", "rawMarkdown": "It's the number of ml you think you might be off regarding your prediction", "votes": null }, { "id": "988732", "postDate": "08/28/2020 08:32:03", "content": "Confidence is a confusing name because lower values are better. It should be called uncertainty (or something like that). \nIt has the same units as the predicted FVC (ml) because it represents the range of FVC values you expect to see.
\n\nThe minimum uncertainty (i.e. confidence) is 70 ml because that is the accuracy of the spirometer they use to take the measurment (I also suspect it is because a value of 0 would break the evaluation metric).
\n\n
Strictly speaking, it is √2 times the number of ml you think you might be off due to parametrization of the Laplace distribution used in this competition.
", "rawMarkdown": "Strictly speaking, it is √2 times the number of ml you think you might be off due to parametrization of the Laplace distribution used in this competition.", "votes": null }, { "id": "988799", "postDate": "08/28/2020 09:35:54", "content": "Isn't that factor multiplied after the value you input?
", "rawMarkdown": "Isn't that factor multiplied after the value you input?", "votes": null }, { "id": "988850", "postDate": "08/28/2020 10:27:01", "content": "No, we're supposed to predict the standard deviation so you have to multiply it before.
", "rawMarkdown": "No, we're supposed to predict the standard deviation so you have to multiply it before.", "votes": null } ], "comments": [ { "id": 987673, "author_name": "lukereijnen", "author_url": "", "post_date": "08/27/2020 12:35:08", "content": "It's the number of ml you think you might be off regarding your prediction
", "votes": null, "replies": [ { "id": 988750, "author_name": "christoffer", "author_url": "", "post_date": "08/28/2020 08:44:25", "content": "Strictly speaking, it is √2 times the number of ml you think you might be off due to parametrization of the Laplace distribution used in this competition.
", "votes": null, "replies": [] }, { "id": 988799, "author_name": "lukereijnen", "author_url": "", "post_date": "08/28/2020 09:35:54", "content": "Isn't that factor multiplied after the value you input?
", "votes": null, "replies": [] }, { "id": 988850, "author_name": "christoffer", "author_url": "", "post_date": "08/28/2020 10:27:01", "content": "No, we're supposed to predict the standard deviation so you have to multiply it before.
", "votes": null, "replies": [] } ] }, { "id": 988732, "author_name": "samsammurphy", "author_url": "", "post_date": "08/28/2020 08:32:03", "content": "Confidence is a confusing name because lower values are better. It should be called uncertainty (or something like that). \nIt has the same units as the predicted FVC (ml) because it represents the range of FVC values you expect to see.
\n\nThe minimum uncertainty (i.e. confidence) is 70 ml because that is the accuracy of the spirometer they use to take the measurment (I also suspect it is because a value of 0 would break the evaluation metric).
\n\n