Hi Nicolas,
When you say you "sorted" the data into columns, did you use the sort operation?
If you sorted each block of data into ascending or descending order, then those sorted blocks would be highly correlated. That correlation would be an expected outcome of the sort.
cheers
Jamie
________________________________________
From: FSL - FMRIB's Software Library <[log in to unmask]> on behalf of Nicolas J. Bourguignon <[log in to unmask]>
Sent: 27 January 2016 15:47
To: [log in to unmask]
Subject: [FSL] Near-perfect correlation in times series
Hello FSL folks,
I was wondering about a phenomenon that I just noticed and which worries me a bit. I was extracting individual subjects' time-series from ROIs using fslmeants with the goal of running ANOVAs. The input files from which I got the time-series was filtered_func_data. The time-series were organized into blocks of three speech production tasks (30 time-points per task). After sorting task-related time-points into different columns in Excel and running a simple correlation, I noticed a near-perfect correlation between them (I mean, 0.999998).
Does anyone have an explanation for this? Is this normal?
Thanks!
Nicolas
|