Back to PLS HelpPosted on 12/15/17 06:13:19
Number of posts: 13
Hi there,
I've conducted a seed PLS analysis and would like to plot each condition's correlation score across each lag of my task (i.e. the temporal brain score plots for each conditon collapsed onto one graph). I've extracted the bcorr_data from the gui and obtained a 36 x 7 matrix. The study involves 6 conditions and 7 lags. I'm assuming that the 7 columns refer to lags, but was hoping you could clarify what the 36 columns represent.
Many thanks,
Heather
Replies:
Posted on 12/15/17 07:00:47
Number of posts: 394
Hi Heather
The 36 rows of the matrix will correspond to behavior weights within task. I assume you have 6 behaviors?
Does that sound correct?
Randy
Posted on 12/15/17 09:30:02
Number of posts: 13
Hi Heather
The 36 rows of the matrix will correspond to behavior weights within task. I assume you have 6 behaviors?
Does that sound correct?
Randy
Posted on 12/15/17 10:00:22
Number of posts: 394
You have it correct, Heather. Go for it!
cheers
Randy
Posted on 01/12/18 08:52:08
Number of posts: 13
You have it correct, Heather. Go for it!
cheers
Randy
Hi again Randy,
Thank you again for all the help, it's much appreciated. Just a couple quick follow up questions for you to make sure I understand properly.
The multiple voxels extraction takes the seed voxel's activity across each time lag (7 in my study) in each condition, yielding a results file with 6 values per condition. Why is this?
Secondly, when plotting error bars in the temporal brain correlations plot, is it alright to use the standard error or are there custom values stored in a results file I should be using instead?
Lastly, when comparing the temporal brain correlation plot to the correlation overview plot I'm noticing that the values follow the same rough general pattern, but don't match. Is this because of bootstrapping or mean-centring of the data in the latter, but not the former, case? Or is there something else I'm missing?
Thanks again!
Best,
Heather
Posted on 01/14/18 15:51:20
Number of posts: 394
Hi Heather - see below
The multiple voxels extraction takes the seed voxel's activity across each time lag (7 in my study) in each condition, yielding a results file with 6 values per condition. Why is this?
I *think* it drops the first lag if you use lag0 as the ref scan. Is the first lag in your extracted file zero?
Secondly, when plotting error bars in the temporal brain correlations plot, is it alright to use the standard error or are there custom values stored in a results file I should be using instead?
We don't compute those, unfortunately, as the plot are meant to be descriptive.
Lastly, when comparing the temporal brain correlation plot to the correlation overview plot I'm noticing that the values follow the same rough general pattern, but don't match. Is this because of bootstrapping or mean-centring of the data in the latter, but not the former, case? Or is there something else I'm missing?
I am not sure. Can you either post or send a screen capture?
Login to reply to this topic.
