Parse, filter data and extract x_values

[artainmo]

Parse, filter data and extract x_values

[artainmo]

Use physionet to get the data. Here is the data related to motor movement https://physionet.org/content/eegmmidb/1.0.0/.
Maybe instead of pushing the dataset to github, use MNE-API for physionet or download on local computer via other ways.

Use MNE to parse the data. Tutorial (https://mne.tools/dev/auto_tutorials/index.html).

[artainmo]

Write a script to visualize raw data then filter it to keep only useful frequency bands, and visualize again after this preprocessing.

Get the answers from the raw data annotations. When plotting frequency_bands try to add the answers...

[artainmo]

This part is where you’ll decide which features you’ll extract from the signals.
One extractable feature example is to use the power of the signal by frequency and channels/electrodes.
Another example is to use the signal spectrum. Which can be found using fourier or wavelet transform.
Here is a tutorial that may help with extracting features (https://labeling.ucsd.edu/tutorial/labels).

[artainmo]

Split the datas of each frequency band in the different states (T0,T1,T2).

For each state data extract the:

• power of the signal by frequency and channels/electrodes (PSFC)

Now you can create a dataset with y being equal to states (T0,T1,T2) and x being equal to the following features for each electrode:

• PSFC of delta frequency band (FB)
• PSFC of alpha FB
• PSFC of gamma FB

[artainmo]

Maybe other x values:

• the mean of psd instead of sum...