What’s new?#

0.5 (unreleased)#

The package build backend was switched from setuptools to hatchling. For users, nothing should change, by Richard Höchenberger in #335
autoreject now requires a minimum Python version of 3.11, by Stefan Appelhoff in #339 and Eric Larson in #359
The minimum supported versions of the dependencies were raised to roughly three years old: numpy >= 1.25.2, scipy >= 1.11.2, mne >= 1.6.0, scikit-learn >= 1.3.0, joblib >= 1.3.2 and matplotlib >= 3.7.2, by Eric Larson in #359
autoreject.AutoReject is now about 5x faster, with unchanged results: the cross-validation splits, test-fold medians and train-fold peak-to-peaks are hoisted out of the threshold and parameter searches, the peak-to-peaks needed to pick the channels to interpolate are computed for all epochs at once, and epochs that need the same set of channels interpolated are now repaired together instead of one at a time, by Eric Larson in #359
The channel-level thresholds computed with thresh_method='bayesian_optimization' are now compared on a single cross-validation split rather than on a freshly drawn one per candidate threshold. This only affects random_state=None, where the results were not reproducible to begin with, by Eric Larson in #359

Updates for MNE-Python compatibility by Eric Larson in #324
Fixed bug that arose if picks was specified with an unsorted list, by Paul Bogdan in #312

Updates for MNE-Python compatibility by Eric Larson and Stefan Appelhoff in #302

NumPy 1.24 and Sphinx compatibility updates by Eric Larson in #293 and Stefan Appelhoff in #292

compute_thresholds now works without channel location data with option augment=False by Nikolai Chapochnikov in #282
get_rejection_threshold now also accepts ECoG and SEEG data by Nikolai Chapochnikov in #281
RANSAC implementation was refactored, works now with n_jobs>1 and produces consistent results across different number of jobs. Testing on simulated data added. by Simon Kern in #280
autoreject now requires mne >= 1.0, by Mainak Jas, Alex Gramfort, and Stefan Appelhoff in #267 and #268
Add reject_log option to autoreject.AutoReject.transform() to enable users to make corrections to reject_log estimated by autoreject, by Alex Rockhill in #270
Add autoreject.RejectLog.save() and autoreject.read_reject_log() to save and load reject logs, by Alex Rockhill in #270

Make plotting of reject log more flexible by allowing to change the orientation of plot, option to supress the immediate display of the plot and returning the plot figure for further modification or saving in autoreject.RejectLog.plot(), by Marcin Koculak in #152
Add show_names option to autoreject.RejectLog.plot(), by Mainak Jas in #209
Enable support for fNIRS data types, by Robert Luke in #177
Add additional type support for argument picks, by Mathieu Scheltienne in #225
Use MNE progressbar, by Patrick Stetz in #227
autoreject.RejectLog.plot() now accepts an ax parameter to pass an existing Axes object, by Stefan Appelhoff in #301

Don’t reset epochs.info[‘bads’] within autoreject.compute_thresholds(), by Mainak Jas in #203
Adjust usage of a private MNE method used for interpolation that failed for newer MNE versions, by Adina Wagner in #212.

Introduced a new method autoreject.AutoReject.save() and function autoreject.read_auto_reject() for IO of autoreject objects, by Mainak Jas in #120
Make MEG interpolation faster by precomputing dot products for the interpolation, by Mainak Jas in #122
Add default option for param_range in autoreject.validation_curve(), by Alex Gramfort in #129

Fixed bug in picking bad channels during interpolation. This bug only affects users who got an assertion error when running autoreject.AutoReject, by Mainak Jas in #115
Added check for channel locations so that autoreject does not hang when the channel positions are nan, by Mainak Jas in #130
Fixed bug in random seed for the Ransac algorithm when n_jobs > 1, by Legrand Nico and Mainak Jas in #138
Fixed pickling of autoreject.AutoReject, by Hubert Banville in #193

Added ch_types argument to autoreject.get_rejection_threshold() to find rejection thresholds for only subset of channel types in the data, by Mainak Jas in #140