Hello,
I'm trying to run context-dependent correlation analysis, and at the stage where I want to remove the trend from my seed timeseries (listed as step 2 in Gang's notes).
There seems to be two ways to do this: 3dDetrend and 3dSynthesize
I used the following for 3dDetrend (exactly from Gang's):
3dDetrend -polort 2 -prefix SeedR Seed.1D\'
1dtranspose SeedR.1D Seed_ts.1D
For the 3dSynthesize option:
3dSynthesize -cbucket cbucket+orig -matrix x1D.x1D select baseline allfunc -prefix effectofnointerest
3dcalc -a input+orig -b effectofnointerest -expr 'a-b' -prefix cleandata
I then extracted the Seed.1D from cleandata dataset.
Attached are sample results of these two options; however, they seem to yield quite different results.
My questions are 1) are these two ways of removing trend/effectofnointerest supposed to yield pretty similar results? (i.e., 1D output vales), 2) which one is theoretically more sound or yielding more accurate measure?
Thank you very much for your help!
Jinny