Hello,\nI'm newbie to datascience,I have got the same prediction value for each image in my test data..\nwhat should be the reason? I have tried many models but in final prediction, got the same value for each image. Is anybody facing the same problem?
", "messages": [ { "id": 913489, "postDate": "2020-07-03T08:39:29.143Z", "content": "Hello,\nI'm newbie to datascience,I have got the same prediction value for each image in my test data..\nwhat should be the reason? I have tried many models but in final prediction, got the same value for each image. Is anybody facing the same problem?
", "rawMarkdown": "Hello,\nI'm newbie to datascience,I have got the same prediction value for each image in my test data..\nwhat should be the reason? I have tried many models but in final prediction, got the same value for each image. Is anybody facing the same problem?", "votes": 1 }, { "id": 913643, "postDate": "2020-07-03T10:31:55.547Z", "content": "I got these error. But I need to see your code or just a put a link to your notebook here
", "rawMarkdown": "I got these error. But I need to see your code or just a put a link to your notebook here", "replies": [ { "id": 914736, "postDate": "2020-07-04T07:29:46.470Z", "rawMarkdown": "", "isDeleted": true }, { "id": 914742, "postDate": "2020-07-04T07:36:16.180Z", "content": "Here is my kernel link:\nhttps://www.kaggle.com/anuragkushwah/kernel15cd6cce67
\n\nEach test image got the same prediction value. Also the validtion accuracy is also same for every epoch.
", "rawMarkdown": "Here is my kernel link:\nhttps://www.kaggle.com/anuragkushwah/kernel15cd6cce67\n\nEach test image got the same prediction value. Also the validtion accuracy is also same for every epoch." }, { "id": 915042, "postDate": "2020-07-04T12:46:04.820Z", "content": "404 error
", "rawMarkdown": "404 error" }, { "id": 915066, "postDate": "2020-07-04T13:01:30.810Z", "content": "@anuragkushwah Your kernel is Private. You need to make it Public.
", "rawMarkdown": "@anuragkushwah Your [kernel](https://www.kaggle.com/anuragkushwah/kernel15cd6cce67) is Private. You need to make it Public." }, { "id": 915251, "postDate": "2020-07-04T15:29:37.510Z", "content": "Sorry guys! @msharuk589 @sirishks plzz check it now.
", "rawMarkdown": "Sorry guys! @msharuk589 @sirishks plzz check it now." } ] }, { "id": 913636, "postDate": "2020-07-03T10:24:33.407Z", "content": "Hello, \nwithout piece of your code - hard to tell.\nIn particular in this competition, there is a lot of data with imbalances (approx 1.76% of total images have target = 1). Hence you can train a model that will only predict target = 0 (no melanoma) with a pretty high accuracy, but auc (competiton metric) will be about 0.5.
", "rawMarkdown": "Hello, \nwithout piece of your code - hard to tell.\nIn particular in this competition, there is a lot of data with imbalances (approx 1.76% of total images have target = 1). Hence you can train a model that will only predict target = 0 (no melanoma) with a pretty high accuracy, but auc (competiton metric) will be about 0.5. " }, { "id": 913535, "postDate": "2020-07-03T09:09:17.940Z", "content": "Are you using a public kernel? If not, start with one of those notebooks which already works correctly end-to-end.
\n\nAlso, you are expected to provide a link to your kernel so that others can provide feedback.
\n\nPS: Perhaps questions such as these should be posted in the official Topic New to Machine Learning or Kaggle?
", "rawMarkdown": "Are you using a [public kernel](https://www.kaggle.com/c/siim-isic-melanoma-classification/notebooks?sortBy=voteCount&group=everyone&pageSize=20&competitionId=20270)? If not, start with one of those notebooks which already works correctly end-to-end.\n\nAlso, you are expected to provide a link to your kernel so that others can provide feedback.\n\nPS: Perhaps questions such as these should be posted in the official Topic [New to Machine Learning or Kaggle?](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/154297)" } ], "comments": [ { "id": 913643, "author_name": "MhdSharuk", "author_url": "", "post_date": "2020-07-03T10:31:55.547000", "content": "I got these error. But I need to see your code or just a put a link to your notebook here
", "votes": 0, "replies": [ { "id": 914736, "author_name": "", "author_url": "", "post_date": "2020-07-04T07:29:46.470000", "content": "", "votes": 0, "replies": [] }, { "id": 914742, "author_name": "Mr_Novice", "author_url": "", "post_date": "2020-07-04T07:36:16.180000", "content": "Here is my kernel link:\nhttps://www.kaggle.com/anuragkushwah/kernel15cd6cce67
\n\nEach test image got the same prediction value. Also the validtion accuracy is also same for every epoch.
", "votes": 0, "replies": [] }, { "id": 915042, "author_name": "MhdSharuk", "author_url": "", "post_date": "2020-07-04T12:46:04.820000", "content": "404 error
", "votes": 0, "replies": [] }, { "id": 915066, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2020-07-04T13:01:30.810000", "content": "@anuragkushwah Your kernel is Private. You need to make it Public.
", "votes": 0, "replies": [] }, { "id": 915251, "author_name": "Mr_Novice", "author_url": "", "post_date": "2020-07-04T15:29:37.510000", "content": "Sorry guys! @msharuk589 @sirishks plzz check it now.
", "votes": 0, "replies": [] } ] }, { "id": 913636, "author_name": "Wojtek Rosa", "author_url": "", "post_date": "2020-07-03T10:24:33.407000", "content": "Hello, \nwithout piece of your code - hard to tell.\nIn particular in this competition, there is a lot of data with imbalances (approx 1.76% of total images have target = 1). Hence you can train a model that will only predict target = 0 (no melanoma) with a pretty high accuracy, but auc (competiton metric) will be about 0.5.
", "votes": 0, "replies": [] }, { "id": 913535, "author_name": "Sirish Somanchi", "author_url": "", "post_date": "2020-07-03T09:09:17.940000", "content": "Are you using a public kernel? If not, start with one of those notebooks which already works correctly end-to-end.
\n\nAlso, you are expected to provide a link to your kernel so that others can provide feedback.
\n\nPS: Perhaps questions such as these should be posted in the official Topic New to Machine Learning or Kaggle?
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "913489": "Hello,\nI'm newbie to datascience,I have got the same prediction value for each image in my test data..\nwhat should be the reason? I have tried many models but in final prediction, got the same value for each image. Is anybody facing the same problem?", "913643": "I got these error. But I need to see your code or just a put a link to your notebook here", "913636": "Hello, \nwithout piece of your code - hard to tell.\nIn particular in this competition, there is a lot of data with imbalances (approx 1.76% of total images have target = 1). Hence you can train a model that will only predict target = 0 (no melanoma) with a pretty high accuracy, but auc (competiton metric) will be about 0.5. ", "913535": "Are you using a [public kernel](https://www.kaggle.com/c/siim-isic-melanoma-classification/notebooks?sortBy=voteCount&group=everyone&pageSize=20&competitionId=20270)? If not, start with one of those notebooks which already works correctly end-to-end.\n\nAlso, you are expected to provide a link to your kernel so that others can provide feedback.\n\nPS: Perhaps questions such as these should be posted in the official Topic [New to Machine Learning or Kaggle?](https://www.kaggle.com/c/siim-isic-melanoma-classification/discussion/154297)" } }