This data includes EEG recordings from 4 subjects (3 train and 1 test subjects) monitored during a night of slip with hand-annotations for sleep-related markers. The task is to predict the markers from the EEG data. 

We provide the **raw**, not organized into epochs, EEG data (`.mat` files) and labels (`.csv`) files for all three test subject, and only the EEG data for the fourth. For convenience, we also provide data organized in **epochs** (30s chunks of data aligned to each event onset with the associated marker; in `.npy` files). Participants are encouraged to use the raw un-cut data, apply their own de-noising and pre-processing steps to it and then organizing the data into epochs for training the models and preparing the submission. This is not a requirement, if preferred the pre-processed data made available here can be used directly. 
We provide [detailed instructions](https://github.com/BCI-I/BCII-IDUN-Challenge-tutorials)  on how to process raw EEG and labels data into epochs. These should be followed closely to ensure that the data is correctly formatted for submission. 

More **details about the files**:
- `.mat` files: raw EEG data (of shape `1 x n_samples`). One for each subject, 4 in total.
- `.csv` files: raw marker labels. Columns include `Marker` and `Timestamp_samples`. 3 in total (test subject excluded)
- `subject_X.npy`: EEG data organized with 30s epochs. Of shape `n_epochs x epoch_length`. (4 in total)
- `subject_Y.npy`: labels for each epoch. Of shape `n_epochs` with the markers names in it. (3 in total). 

The subjects are: `train=["S001", "S002", "S003"]` and `test=["S004"]`. 

We provide [detailed code](https://github.com/BCI-I/BCII-IDUN-Challenge-tutorials) to load and processed the raw data into epochs as well as to use epochs for training and evaluating ML models and preparing a submission.