Unfold v1.3.1

Bugfix-release

Fixed a bug in cyclical splines: the wrapping code was not working correctly when using quantile-placement. e.g. if the limits where 0, 360 - but 520 was inputted, this was wrongly mapped (due to discrepancy of max knot & bound)

Full Changelog: 1.3...1.3.1

Unfold v1.3

Summary 1.3

some minor fixes that should be put in a release.

Breaking (minor) default change

• we switched the ASR "detect bad data" from 1e-10 to 1e-5 - thus it removes LESS data. We found it in multiple dataset to be more appropriate.

What's Changed

• silenced uf_epoch output of eeglab by @behinger in #112
• Small changes to make lines more visible in design matrix (Xdc) by @behinger in #113
• docu fix ASR (thanks to @annakau 🎉) #135
• Github CI for automatic unittests

New Contributors

• @annakau made their first contribution in #135 - thanks a ton!

Full Changelog: 1.2...1.3

Unfold1.2

Summary:

• Lot's of bugfixes
• New uf_reject functions. These are based on ASR and jointprob, partially contributed by @wanjam. Thanks!
  
• Improved documentation

Details

• fixed bug where eeglab was loaded too late
• improved uf_checkmodelfit for (crossvalidated)(partial)R² measures
• fixed a bug in uf_erpimage where two events might happen at the same time
• fixed a str-bug. Thanks @nicolaslanger
• fixed an error message in uf_glmfit
• better warning messages in uf_glmfit
• uf_glmfit: reduced the necessary tolerance to stop the iterative solver to 10^-8
• uf_glmfit_nodc: fix cvglmnet
• uf_plot_2nd: added robust second stage measure (trimmed mean)
• uf_plotDesignmat: define how many seconds are plotted
• uf_predictContinuous: fix for TRF
• added a bunch of unit-tests

Unfold 1.1

Unfold 1.1

Features

single trial erpimages

Calculate ERPimages as seen in our publications. These erpimages can be of raw data, modelled data or of the residuals. Options to additionally include residual activity into the modelled is available. In addition one can remove/keep predictors from the modelled/residual output. This is helpful to e.g. simulate what it would look like to keep eventA, but remove eventB. Or to remove e.g. the effect of saccadeAmplitude but keep the effect of FixationPosition etc. See the new tutorial for more usage cases

Temporal response functions (TRFs)

TRFs allow to estimate brain responses from time-continuous predictors. E.g. if you record an audiostream concurrently with the EEG signal, you could include it as a timeexpanded predictorset. You could also use it for features of audio, pupil, continuous stimulation streams etc. See the tutorial for a usage example.

R² and partial R² (crossvalidated)

calculate explained variability of the model. Total R² over all modeled data and partial R² (totalR² - R²withoutPredictor), both with or without crossvalidation. Currently it is not possible to calculate R² for single time-points, but only for the complete modelled data.

uf_addmarginal

Addmarginal now supports marginal effect at the mean (MEM) and average marginal effect (AME). In short it is the difference of E[f(x)] and f(E[x]). This is the same for continuous variables but not for splines. Starting from Unfold 2.0 we will set the default to AME. For backward compatibility we will keep it by default at MEM

3 new tutorials

Tutorials on the peerJ paper, on erpimages, on temporal response functions

Bugfixes

• Major: uf_continuousArtifactDetect.m had a bug inherited from ERPlab. For artefact detection only the last channel was used!! This is now fixed.
• init_unfold now uses eeglab redraw instead of eeglab, not overwriting an already loaded ÈEG structure.
• interactions with more than two levels had a bug while renaming, only if they occured in multiple events.
• several smaller bugfixes

Unfold 1.0

With publication of our peer-reviewed journal article, we have some new features, but mostly included some bugfixes and additional documentation. Even though we change the version number, all changes are 100% backward-compatible.

In general, though, we recommend using the "git-master" branch to stay up-to-date.

Initial Release

This is the inital release for the second revision of the unfold-toolbox bioRxiv preprint.

Once the toolbox reference paper is peer-reviewed and the toolbox further tested, we will reach v1.0