![](\"https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F7223266%2F1ab03a965b1b74d78d0488f42cc6d528%2Fdownload.png?generation=1705320165568682&alt=media\")
\nThanks for your help !
Hello, i'm stuck while trying to understand the dataset.
\nThe challenge says that the test_eeg is exactly 50 seconds of EEG data but the test_sectrogram is based on 10 minutes of EEG data so how is it possible to know which 50 seconds of the train_eeg to take to make both the same dimensions ?
\nIn addition, the sentence \"The expert annotators reviewed 50 second long EEG samples plus matched spectrograms covering 10 a minute window centered at the same time and labeled the central 10 seconds.\" means that the central 10 seconds of the spectrogram and of the eeg are the same right ? That means that's the same in the test_set even if there is only 1 spectrogram and eeg ?
\n~~~~
\nSo my intuition is that we have to take each 10 central seconds of each sub_eeg on the train_set and aggregate them (like min, mean or other) to have only 1 value per feature (like LL_0.59, LL_0.78, LL_0.98, LL_1.17, LL_1.37) and we can do the same on the test_set to have same dimensions.
\nThanks for your help !
Hello, i'm stuck while trying to understand the dataset.
\nThe challenge says that the test_eeg is exactly 50 seconds of EEG data but the test_sectrogram is based on 10 minutes of EEG data so how is it possible to know which 50 seconds of the train_eeg to take to make both the same dimensions ?
\nIn addition, the sentence \"The expert annotators reviewed 50 second long EEG samples plus matched spectrograms covering 10 a minute window centered at the same time and labeled the central 10 seconds.\" means that the central 10 seconds of the spectrogram and of the eeg are the same right ? That means that's the same in the test_set even if there is only 1 spectrogram and eeg ?
\n~~~~
\nSo my intuition is that we have to take each 10 central seconds of each sub_eeg on the train_set and aggregate them (like min, mean or other) to have only 1 value per feature (like LL_0.59, LL_0.78, LL_0.98, LL_1.17, LL_1.37) and we can do the same on the test_set to have same dimensions.
\nThanks for your help !
@djanyigit go through Discussion well explained
", "rawMarkdown": "djanyigit go through [Discussion](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010) well explained", "votes": null } ], "comments": [ { "id": 2602955, "author_name": "seshurajup", "author_url": "", "post_date": "01/15/2024 13:19:02", "content": "@djanyigit go through Discussion well explained
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "2602901": "Hello, i'm stuck while trying to understand the dataset.\n\nThe challenge says that the test_eeg is exactly 50 seconds of EEG data but the test_sectrogram is based on 10 minutes of EEG data so how is it possible to know which 50 seconds of the train_eeg to take to make both the same dimensions ?\n\nIn addition, the sentence \"The expert annotators reviewed 50 second long EEG samples plus matched spectrograms covering 10 a minute window centered at the same time and labeled the central 10 seconds.\" means that the central 10 seconds of the spectrogram and of the eeg are the same right ? That means that's the same in the test_set even if there is only 1 spectrogram and eeg ?\n~~~~\nSo my intuition is that we have to take each 10 central seconds of each sub_eeg on the train_set and aggregate them (like min, mean or other) to have only 1 value per feature (like LL_0.59, LL_0.78, LL_0.98, LL_1.17, LL_1.37) and we can do the same on the test_set to have same dimensions.\n\n\nThanks for your help !", "2602955": "djanyigit go through [Discussion](https://www.kaggle.com/competitions/hms-harmful-brain-activity-classification/discussion/468010) well explained" }, "source": "meta" }