Is there a way to do bootstrapping on epochs to have different evoke ? iter_evoked()

Support & Discussions
, ,
1

Hello everyone,

I am trying to implement Standardized Measurement Error (Luck et al., 2021) which has a step of doing bootstrapping.

It consists of selecting randomly with replacement a set of n epochs and doing the average on these to have slightly different evoked on which one can calculate peak amplitude and latenciy to estimate the SME.

So I am trying to randomly select n epochs from an evoked object but I have difficulties with using iter_evoked() which iterate sequentially through the epochs of the evoked object (thus disabling random sampling).

More, I am not able to apply average() on the epochs returned by iter_evoked() to constitute new evoked and if I assign the individual epochs retrieved by iter_evoked() to a list I can’t do the average neither.

I want to avoid accessing the data through get_data() as I working with evoked object has the advantage of allowing different manipulations without accessing data directly.

Any help, comment or advice will be greatly appreciated.

Thank you very much.

2

You can index the epochs object directly:

random_picks = [4, 6, 10, 15, 14]
evoked = epochs[random_picks].average()
3

Thanks you very much for this solution!