----------------------
This release adds two EXPERIMENTAL flags:
--normalize-betas and --no-signal-scaling.
I'm especially unsure about the behavior of
--normalize-betas in combination with any hrf that contains
"derivatives" or "derivatives + dispersion", since I'm unsure
when to divide the the beta estimates with the variances of
the beta estimates to derive the t-statistic
(i.e., normalizing the beta estimates).
This release also has given access to residuals from the LSA and LSS
interfaces.
Another WARNING: residuals for the LSS interface are not quite as meaningful
since I'm averaging the residuals over all the models being generated.
This project is not version 1.x.x yet, so be aware aspects of the API
may change rapidly.
Thanks to michaelmack for opening an issue to suggest new features and
PeerHerholz for their contributions!
* [ENH] add --normalize-betas option (299) jdkent
* [ENH] add --no-signal-scaling option (298) jdkent
* [ENH,FIX] optimally combine temporal + dispersion derivatives for beta estimation (296) jdkent
* [ENH] add residuals to LSS/LSA interfaces (294) jdkent
* [FIX] remove standardize parameter (292) jdkent
* [ENH] Allow niftiimage input (289) jdkent
* fix singularity install instructions in docs (288) PeerHerholz