new_y in extract function assigning different label which mislead the model

[manoji-s]

In extract function from BCI_functions.ipynb, if the averaged y label of a particular window is a float (some y labels is from one class and remaining from another class during transition), the y label assigned to that window is 0. Instead it should be either floor or ceil of that averaged value right?

E.g: Consider y values of a window as [2,2,2,2,2,2,2,3,3,3,3,3,3,3,3,3] => ave_y = 2.56
Either this window's y label should be 2 or 3.
But this function will assign it as 0 (in the else part)

Am I miss something here? Please clarify me this. Thanks!

[ShieldY5]

When I try to remove tags other than 4 and 5, I encountered the same problem. The tags of training set and verification set are all 0