Fitlins

------------------------

New feature release in the 0.8.x series.

This release includes a change where events that are not explicitly convolved
are modeled with boxcar functions instead of the nistats default.

* FIX: Sparse matrix should be in long format, and HRF should be None by default (https://github.com/poldracklab/fitlins/pull/259)

------------------------

Bug-fix release in the 0.7.x series.

This release requires PyBIDS 0.12.2 or higher, fixing various bugs that affected FitLins.

* Only drop modulation if needed (https://github.com/poldracklab/fitlins/pull/253)

-----------------------

New feature release in the 0.7.x series. This is an accumulation of changes over
several months.

This release requires PyBIDS 0.12 and includes some *quite provisional* CIFTI-2 support
as well as log-likelihood and R^2 maps for assessing goodness of fit.

* FIX: Downgrade Sphinx to <3.0 (https://github.com/poldracklab/fitlins/pull/226)
* FIX: Validate BIDS datasets except when searching for models (https://github.com/poldracklab/fitlins/pull/214)
* FIX: Blacklist coverage >=5 (https://github.com/poldracklab/fitlins/pull/209)
* ENH: Bundle Conte69 surfaces (https://github.com/poldracklab/fitlins/pull/249)
* ENH: Add CIFTI-2 support (https://github.com/poldracklab/fitlins/pull/239)
* ENH: Add logL as model output (https://github.com/poldracklab/fitlins/pull/230)
* ENH: Save model level output (r-squared) (https://github.com/poldracklab/fitlins/pull/224)
* ENH: Add FEMA as key value to stat `path_pattern` (https://github.com/poldracklab/fitlins/pull/220)
* RF: Accept nipype configurations from users (https://github.com/poldracklab/fitlins/pull/229)
* MNT: Update requirements (https://github.com/poldracklab/fitlins/pull/225)
* MNT: Bump pybids to 0.10.2 (https://github.com/poldracklab/fitlins/pull/216)
* CI: Improve outputs (https://github.com/poldracklab/fitlins/pull/218)

-------------------------

Hotfix release.

* FIX: FEMA input index should filter on rows, not columns. (https://github.com/poldracklab/fitlins/pull/207)

-------------------------

Hotfix release.

* FIX: Add desc=preproc as filter when finding preproc BOLD files (https://github.com/poldracklab/fitlins/pull/204)

-------------------------

New feature release in the 0.6.x series.

This release respects recent changes to the BIDS-StatsModels draft
specification to support fixed-effects meta-analysis (FEMA) contrasts,
and renames "AutoContrasts" to "DummyContrasts".

Provisional support for F-tests has been added.

Additional rearchitecting by Dylan Nielson provides significant speedups
for large datasets by caching BIDS layout information.

* SPEC: Add fixed effects (FEMA) contrasts (https://github.com/poldracklab/fitlins/pull/191)
* SPEC: Change AutoContrasts to DummyContrasts (https://github.com/poldracklab/fitlins/pull/197)
* FIX: Don't pass ignore and force index to `init_fitlins_wf` (https://github.com/poldracklab/fitlins/pull/202)
* FIX: BIDSelect confusion between Nifti and JSON (https://github.com/poldracklab/fitlins/pull/193)
* FIX: Set `smoothing_fwhm` after creating next level. (https://github.com/poldracklab/fitlins/pull/190)
* FIX: Recognize cosine columns without underscores (https://github.com/poldracklab/fitlins/pull/185)
* ENH: Update logging levels (https://github.com/poldracklab/fitlins/pull/198)
* ENH: Add option to load BIDS layouts from database file (https://github.com/poldracklab/fitlins/pull/187)
* ENH: Add memory estimate for first-level models, enable memory management (https://github.com/poldracklab/fitlins/pull/199)
* ENH: Avoid casting BOLD data to float64 if possible (https://github.com/poldracklab/fitlins/pull/196)
* ENH: Add F-tests (https://github.com/poldracklab/fitlins/pull/195)
* ENH: Drop missing model inputs (https://github.com/poldracklab/fitlins/pull/183)
* RF: Abstract interfaces to simplify swappability (https://github.com/poldracklab/fitlins/pull/188)