v0.4.4

New Features

• Functions frites.conn.conn_reshape_undirected and frites.conn.conn_reshape_directed now supports multidimensional arrays (21ba7e0)
• New function frites.core.ent_nd_g to compute entropy on tensors (17587a1)
• New function frites.conn.conn_ii to estimate the interaction information (10938b4)
• New function frites.conn.conn_pid to estimate the partial information decomposition (ac9798d)

• New function frites.conn.conn_fit to estimate the feature specific information transfer (#59) - @aopy

Bug fixes

• Base 2 frites.core.ent_nd_g (686854d)

New Features

• frites.conn.conn_links now accepts to use brain region names as source or target seeds (8d7e1fb)
• New function frites.stats.confidence_interval for computing confidence intervals, standard deviation (sd) and standard error on the mean (sem) on numpy arrays and xarray DataArray (3a51253)
• New plotting module with two functions for plotting connectivity matrices, frites.plot.plot_conn_heatmap and frites.plot.plot_conn_circle (7a107d1, a480371, 33930cc, 24029c4)
• Add the possibility to control the node's size in frites.plot.plot_conn_circle (7454376)
• Add support for signed connectivity and improve node control of frites.plot.plot_conn_circle (78bbb91)

Bug fixes

• Fix parallel computing of frites.conn.conn_covgc (f2ff5f6)
• Use linearly spaced indices instead of closest time points in frites.conn.define_windows (b5443ac)
• frites.conn.conn_spec average over tapers after computing connectivity metric (10966a1).Thanks to @adam2392 and @ruuskas from the mne-connectivity package.

New Features

• New function :func:frites.simulations.sim_ground_truth for simulating spatio-temporal ground-truths (ba44a42)
• New function :func:frites.conn.conn_spec for computing the single-trial spectral connectivity (8151486) - by @ViniciusLima94
• New method :class:frites.workflow.WfMi.confidence_interval method to estimate the confidence interval (ad03919, 8189622, fc58475, fc58475)
• New function :func:frites.conn.conn_net for computing the net connectivity (c86b19f)
• New function :func:frites.set_mpl_style for example styles
• New function :func:frites.conn.conn_links for generating connectivity links (a0d0182)
• New function :func:frites.utils.downsample for down-sampling DataArray (59e15e6)
• :class:frites.estimator.CorrEstimator can no be defined with Pearson or Spearman correlation with both vector or tensor-based implementations (f6e80e9)
• :class:frites.workflow.WfStats now allows to pass rfx_center parameter for controlling whether effect-sizes should be centered and rfx_sigma for the hat correction (aab68e1)
• :func:frites.conn.conn_links can now be used for selecting intra and / or inter-hemispheric connections (87c762c)
• :func:frites.conn.conn_links allows selecting links with inter / intra / both roi connections (74d2cc1)

Internal changes

• Do not allow anymore to add new keys to the CONFIG dict (abea667)

New Features

• New :class:frites.estimator.CustomEstimator for defining custom estimators (e473c71, 5584654)
• New function :func:frites.conn.conn_fcd_corr for computing the temporal correlation across networks (2001f0c)
• New function :func:frites.utils.acf for computing the auto-correlation (48ef0a0)
• New function :func:frites.conn.conn_ccf for computing the cross-correlation (43fceb0)

Bug fixes

• Fix attribute conversion in connectivity functions (b990c76)

New Features

• New estimators (frites.estimator.CorrEstimator, frites.estimator.DcorrEstimator) for continuous / continuous relationships (73ed8bb, bc370a9, cf7a345)
• frites.conn.conn_dfc supports passing other estimators (a864a7b)
• frites.utils.time_to_sample and frites.utils.get_closest_sample conversion functions (7c44478)
• frites.conn.conn_ravel_directed reshaping function (f9b9d27)
• new frites.workflow.WfMi.copy for internal workflow copy (0c2228c, 860f3d4)
• new frites.workflow.WfMiCombine and example class for combining workflows (62072ee)
• new classfrites.estimator.ResamplingEstimator trial-resampling estimator (13f6271)

Bug fixes

• Fix frites.workflow.WfMi.get_params when returning a single output (3bde82e)
• Improve attributes conversion for saving netcdf files (bool and dict) (8e7dddb, c6f7a4d)
• Fix Numpy np.float and np.int warnings related (896a198, 7f2a1ca, 0d1a122)

New Features

• :py:frites.conn.conn_dfc supports multivariate data + improve computing efficiency (1aed842, c4ac490)
• Reshaping connectivity arrays support elements on the diagonal + internal drop of duplicated elements (daac241)
• :py:frites.conn.conn_dfc supports better channel aggregation (a66faa7)

Internal Changes

• Connectivity metric now use the :py:class:frites.dataset.SubjectEphy for internal conversion of the input data
• :py:class:frites.workflow.WfMi.get_params returns single-subject MI and permutations with dimension name 'subject' (instead of subjects) (85884f3)
• All connectivity metrics now use :py:frites.conn.conn_io to convert inputs into a similar format
• Improve CI (#18)

Bug fixes

• Fix :py:class:frites.dataset.SubjectEphy when the data contains a single time point (a33d443)
• Fix attributes of :py:class:frites.conn.conn_covgc (c120626)
• Fix :py:class:frites.dataset.DatasetEphy representation without data copy + html representation (b3ae7b8, #16)
• Fix passing tail input to the :py:class:frites.workflow.WfMi (6df86d1)

New Features

• new :py:class:frites.io.Attributes class for managing and printing datasets' and workflow's attributes (be046b1)
• new :py:class:frites.dataset.SubjectEphy single-subject container (ac22cf4)
• new estimators of mutual-information, :py:class:frites.estimator.GCMIEstimator (901b3cb, 65d1e08, 0015bf5, beed6a0), :py:class:frites.estimator.BinMIEstimator (beed6a0)
• new kernel smoothing function :py:frites.utils.kernel_smoothing

Internal Changes

• Removed files (cdff9b4, 9e96f8e, 14961aa)
• :py:class:frites.dataset.DatasetEphy don't perform internal data copy when getting the data in a specific ROI (2da73ef)
• Compatibility of MI estimators with workflows (7dc76ee, e7a9c23)
• Improve the way to manage pairs of brain regions (8b955a1, bfdf2db, 57c1e4b, b1ff8c3)

Breaking changes

• :py:class:frites.dataset.SubjectEphy and :py:class:frites.dataset.DatasetEphy to specify whether channels should be aggregated (default agg_ch=True) or not (agg=False) when computing MI. The agg_ch replace sub_roi (18d4e24)
• The workflow WfComod has been renamed :py:class:frites.workflow.WfConnComod (b7b5824)

Bug fixes

• Bug fixing according to the new version of :py:class:frites.dataset.DatasetEphy (1a15e05, 7b83a3d, abd1b28, 70bfefb, 5879950, 66acdf2, 4309be9, 6dc2fbf)