thank you for your answer , when I m trying to apply the cwt on epochs , I face a problem , can you help ?

from scipy import signal

import numpy as np

import matplotlib.pyplot as plt

import pywt

waveletname='db1'

coeff = pywt.wavedec(epochs, waveletname, level = 6)

cA6,cD6,cD5,cD4, cD3, cD2, cD1 = coeff

from scipy import signal

import matplotlib.pyplot as plt

widths = np.arange(1, 31)

cwtmatr = signal.cwt(epochs, signal.ricker, widths)

plt.imshow(cwtmatr, extent = [-1, 1, 31, 1])

plt.show()

i get the error : volume and kernel should have the same dimensionality

I used mne python already , im working on Sleep PolySomnoGraphic Database , I extracted the epochs which are under format (channel, time , event) , when I tried to apply the cwt on it for each channel it gives the error mentioned in previous comment.

Thank you for the big effort

- https://stackoverflow.com/questions/64716806/discrete-wavelet-transform-on-my-raw-eeg-data

this was my question in stack overflow vy the way but no one answered.

Thank you again .

Aparna Sathya Murthy Thank you for the big effort , I tried to understand the code you sent me , you mean we can apply the cwt only on 1D arrays data ? but I wanted to apply it on epochs data , so I can have the signal and event at same time , so I can use it as an input for my classification model .

Thank you again

epoched data are [channel x time x epochs] , I tried to flatten the coefficients and it works can you guide me if it's correct ?

Aparna Sathya Murthy thank you for your help , I do really appreciate .

Thank you Aparna Sathya Murthy , what I'm trying to do is this : divide my signal into epochs , and then each epoch would be filtered into different matrix looking like (channel , time , event) and then apply the cwt on each of this small pieces of data so we can have a time frequency feature extraction for every part .

I'm sorry I didn't get it , what do you mean by each array in Epochs Array ?

Aparna Sathya Murthy I see , I get it now , so to decompose our signal into small 30 s of epochs data it should be in array values

we 're planning to work with 2 channels only (EEG channels ) so we will have , (1103, 2,3000) , at first we extract epochs for first raw , from this epochs we extract the events for each 30s of signal , and we apply the cwt on each 3d matrix we got (channel , epoch , event )

Thank you , I ll do the same .

thank you , I have some issues in this part of code , I didn't get what you did exactly thank you for your help

from mne import create_info, EpochsArray

reject = dict(grad=4000)

events = np.empty((40, 3), dtype=int)

event_id = dict(sin50hz=1)

sfreq = 1000.0

n_times = 1024

ch_names = ['SIM0001', 'SIM0002']

ch_types = ['grad', 'grad']

info = mne.create_info(ch_names=ch_names, sfreq=sfreq, ch_types=ch_types)

first_event_sample = 100

for k in range(40):

events[k, :] = first_event_sample + k * n_times, 0, event_id['sin50hz']

thanks for your answer Aparna Sathya Murthy