get mean neighboring channels #508
Conversation
|
|
||
| return NCC, shifts | ||
|
|
||
| def get_neighboring_channel_index(elec_pos, distance=0.75): |
There was a problem hiding this comment.
what do you think about having this function take "ch_idx" as an argument, and only return neighboring_idx[ch_idx], instead of the whole matrix?
i think it might be more useful that way?
probably it will be slower, so in get_mean_neighboring_channels() you might not use this function
| Args: | ||
| data (nch): neural data | ||
| elec_pos (nch, 2): x, y position of each channel | ||
| distance (float): define neighboring channels based on this distance. 0.75 is for 4 neighboring channels and 1.3 is for 9 channels. |
There was a problem hiding this comment.
0.75 and 1.3 only applies to ECoG244 array
|
|
||
| return neighboring_idx | ||
|
|
||
| def get_mean_neighboring_channels(data, elec_pos, distance=0.75): |
There was a problem hiding this comment.
i don't want to derail this function -- it's fine and you can push it -- but just a note for the future: we should really make a spatial smoothing function that takes an arbitrary 2d kernel and smooths the data that way, rather than only doing average and only limited to a radius
| data_test = np.arange(1,10) | ||
| m_data_test = aopy.analysis.get_mean_neighboring_channels(data_test, elec_pos_test, distance=1) | ||
| self.assertTrue(m_data_test[0] == np.sum(data_test[0]+data_test[1]+data_test[3])/3) | ||
| self.assertTrue(m_data_test[4] == np.sum(data_test[1]+data_test[3]+data_test[4]+data_test[5]+data_test[7])/5) |
There was a problem hiding this comment.
it would be great to add a couple more lines to get a spatial map of the test data before and after smoothing and make a plot of before/after. then you could include the plot and the example code in the function docstring as well.
Average ECoG data across neighboring channels based on distance.