v1.8.0

New Features

• Added MultiGroupRCD algorithm, RCD for multiple datasets.
• Add extract_ancestors method to lingam.utils.
• Modified hsic.py to speed up independence test HSIC.
• Added to utils.make_dot the ability to highlight paths between variables in a causal graph.
• Added utils.make_dot_hightlight method to highlight ancestors and descendants of a specified variable in a causal graph.

Experimental features

• Added CausalDataGenerator tool to lingam.experimental.

Code Fixes

• Fixed VAR trend argument used in VARLiNGAM (#86)

Documentation

• Minor text correction on causal_effect.rst

v1.7.1

New Features

• Added CausalBasedSimulator that can generate virtual data using causal graphs.

Code Fixes

• Fixed #58, #70

Tests

• Changed supported python versions

Documentation

• Enhanced tutorials
• Added link to tutorial slides

v1.7.0

New Features

• Added LiM algorithm
• Added CAM-UV algorithm

Code Fixes

• Fixed typos (#50)
• Fixed #44
• Fixed #45
• Fixed #53

v1.6.0

New Features

• Added LiNA and MDLiNA algorithm
• Added RESIT algorithm

v1.5.5

New Features

• Added get_paths method to the result of bootstrapping that searches all paths between specified variables and outputs the bootstrap probability for each path in VARLiNGAM, VARMALiNGAM and LongitudinalLiNGAM

Code Fixes

• Changed the timing for creating the partial orders when applying prior knowledge in DirectLiNGAM
• Fixed that prior knowledge is not applied when there are duplicate no_paths (#34)

v1.5.4

New Features

• Added get_paths method to BootstrapResult that searches all paths between specified variables and outputs the bootstrap probability for each path in ICALiNGAM, DirectLiNGAM, MultiGroupDirectLiNGAM, BottomUpParceLiNGAM and RCD

v1.5.3

New Features

• Changed to apply prior knowledge when pruning of adjacent matrix in DirectLiNGAM and BottomUpParceLiNGAM

Code Fixes

• Corrected issubclass to isinstance in VARLiNGAM and VARMALiNGAM (#25)

v1.5.2

New Features

• Modify algorithm for prior knowledge implemented in DirectLiNGAM and MultiGroupDirectLiNGAM
• Add prior_knowledge option to specify prior knowledge to BottomUpParceLiNGAM

Installation

• Add statsmodels to install_requires in setup.py(#22)

Code Fixes

• Rename the function from BootstrapResult.get_causal_effects to BootstrapResult.get_total_causal_effects
• Change the diagonal component of the prior knowledge matrix created by `utils.make_prior_knowledge from 0 to -1

v1.5.1

• Add get_error_independence_p_values function to all algorithms
• Changed to use Adaptive lasso when estimating the total effect

v1.5.0

• Add RCD algorithm